JP4769586B2 - Plasma reactor and method for improving the uniformity of plasma ion concentration distribution - Google Patents

Description

Cross-reference of related applications

  [01] This application is a continuation-in-part of US Patent Application No. 10 / 841,116 “CAPACTIVELY COUPLED PLASMA REACTOR WITH MANETIC PLASMA CONTROL” filed May 7, 2004 by Daniel Hoffman et al. US Patent Application No. 10 / 192,271, “CAPACTIVELY COUPLED PLASMA REACTOR WITH MAGNETIC PLASMA CONTROL”, filed July 9, 2002 by Hoffman et al., Both of which are assigned to the assignee of the present application.

background

  [02] Capacitively coupled plasma reactors are used to fabricate high aspect ratio semiconductor microelectronic structures. Such structures typically have narrow and deep openings through one or more thin films formed on a semiconductor substrate. Capacitively coupled plasma reactors are used in various processes for making such devices, such as dielectric etch processes, metal etch processes, chemical vapor deposition, and the like. Such a reactor is also used for manufacturing a photolithography mask and a semiconductor flat panel display. Such applications rely on plasma ions that enhance or enable the desired process. The density of semiconductor plasma ions on the surface of the semiconductor workpiece affects process parameters and is particularly important in the fabrication of high aspect ratio microelectronic structures. Indeed, a problem in the fabrication of high aspect ratio microelectronic integrated circuits is that plasma ion concentration non-uniformity across the workpiece surface can lead to process failures due to non-uniform etch or deposition rates. .

  [03] A typical capacitively coupled plasma reactor has a wafer support pedestal in the reactor chamber and a ceiling overlying the wafer support. The sealing can include a gas distribution plate that sprays process gas into the chamber. An RF power source is applied across the wafer support and the ceiling or wall to keep the plasma on the wafer support colliding. Whereas the chamber is generally cylindrical, the sealing and wafer support are rounded coaxially with the cylindrical chamber to enhance uniform processing. Nevertheless, such a reactor has a non-uniform plasma density distribution. Typically, a significant problem is that the radial density distribution of plasma ions is high on the center of the wafer support and low near the periphery. Various approaches to controlling plasma ion concentration distribution have been used to improve process uniformity across the wafer or workpiece surface, at least in part overcoming this problem.

  [04] One such approach is to provide a set of magnetic coils circumferentially spaced around the sides of the reactor chamber and all facing the center of the chamber. A relatively low frequency sinusoidal current is supplied to each coil, but the phase of the sinusoidal current in adjacent coils is offset to generate a slowly rotating magnetic field on the wafer support. This feature tends to improve the radial distribution of plasma ion concentration on the wafer support. When this approach is used in reactive ion etching, it is called magnetic enhanced reactive ion etching (MERIE). This approach has certain limitations. In particular, it may be necessary to limit the strength of the magnetic field in order to avoid device damage to the microelectronic structures on the semiconductor workpiece related to the strength of the magnetic field. The intensity must also be limited to avoid chamber arcing associated with the rate of change of magnetic field strength. As a result, the total MERIE magnetic field may need to be substantially reduced, so substantial limitations can be encountered in plasma ion concentration uniformity control.

  [05] Another approach is called CMF (configurable magnetic field) and uses the same circumferentially spaced coils referred to above. However, in CMF, the coil is operated to apply a magnetic field that extends across the plane of the workpiece support from one side to the other. In addition, the magnetic field rotates about the axis of the wafer support and generates a time averaged magnetic field in the radial direction. In the case of a reactor with four parallel coils, this is all accomplished by applying one DC current to a pair of adjacent coils and a different (reverse) DC current to a pair of adjacent coils on the opposite side. As described above, the coil is switched so that the magnetic field rotates by rotating this pattern. This approach is vulnerable to chamber or wafer arcing by abruptly switching the CMF magnetic field, so the magnetic field strength must be limited. As a result, in some applications, the magnetic field may not be sufficient to compensate for the non-uniformity of the plasma ion concentration generated by the reactor.

  [06] Thus, what is needed is a method to compensate for non-uniformity of plasma ion concentration distribution in a magnetic field more efficiently (so that the magnetic field strength can be lowered) with lower (or no) time variation.

Summary of the Invention

  [07] A plasma reactor for processing a workpiece includes a vacuum chamber defined by sidewalls and a ceiling, a workpiece support surface within the chamber, facing the ceiling, and including a cathode electrode. including. An RF power generator is coupled to the cathode electrode. The plasma distribution is such that the outer annular inner electromagnet in a first plane above the workpiece support surface and the outer annular outer electromagnet in the second plane above the workpiece support surface having a larger diameter than the inner electromagnet. Controlled by an electromagnet and an outer annular bottom magnet in a third plane below the workpiece support surface. A DC current source is connected to each of the inner, outer and bottom electromagnets. The workpiece support pedestal and the inner, outer and bottom magnets can be substantially coaxial. In one embodiment, the first plane is above the second plane, both the first plane and the second plane are above the third plane, and the first, second, and third planes. Is parallel to the workpiece support surface.

[08] The reactor may include processors that control DC current from the inner, outer and bottom DC current sources. Processor, ie
A cusp mode in which a DC current generates an equal and opposite magnetic field at the workpiece support surface in the bottom electromagnet and one of the inner and outer electromagnets;
A mirror mode in which a DC current generates a similar magnetic field at the workpiece support surface in the bottom electromagnet and one of the inner and outer electromagnets;
DC current can be operable in three modes, solenoid mode, in which at least one of the electromagnets generates both a radial and axial magnetic field at the workpiece support surface.

  [09] The processor can improve the plasma ion concentration distribution uniformity with a radial magnetic field at the wafer while improving device damage results by controlling (eg, reducing) the axial magnetic field at the wafer. It can be programmed to search for the best combination of DC currents for a single magnet.

Detailed Description of the Invention

  [67] The plasma ion concentration distribution presented by a particular plasma reactor is a function of chamber pressure, gas mixing and diffusion, and source power radiation pattern. In the reactor of the present invention, this distribution is magnetically modified to approach a selected or ideal distribution that has been predetermined to improve process uniformity. A magnetically altered or corrected plasma ion concentration distribution is one that improves process uniformity across the surface of the wafer or workpiece. To this end, the magnetically corrected plasma distribution can be non-uniform or uniform depending on the needs determined by the user. The inventors have discovered that the average magnetic field strength can improve the efficiency of applying pressure to the plasma to change its distribution to the desired one. This surprising result can be achieved by increasing the radial component of the magnetic field gradient according to this finding. It is understood that the radial direction is approximately the axis of symmetry of the cylindrical chamber. Thus, what is needed is a magnetic field shape with a large radial gradient and a small magnetic field strength in the other direction. Such a magnetic field is a cusp shape whose axis of symmetry coincides with the axis of the cylindrical reactor chamber. One approach to generating a cusp-type magnetic field is to provide coils above and below the cylindrical chamber and to pass DC current through these coils in the opposite direction.

  [68] Depending on the chamber design, it may not be feasible to provide the coil under the wafer pedestal, and therefore, in the first case, the top coil meets these objectives. In addition, what is needed is to make the cusp-type magnetic field configurable or adjustable for precise control or modification of the plasma ion distribution (“ambient” plasma ion distribution) inherent to a given plasma reactor chamber. is there. Such tunability may be essential because the plasma ion distribution provided to different capacitively coupled plasma reactors can vary over a wide range. The radial component of the magnetic field gradient is chosen to apply the magnetic pressure required to change the ambient distribution to the desired distribution. For example, if the desired distribution is a uniform distribution, the applied magnetic field is selected to counteract non-uniformities in the radial distribution of plasma ion concentration presented by the reactor in the absence of the magnetic field. . In this case, for example, if the reactor tends to have a central high distribution of plasma ion concentration, the magnetic field gradient will maintain the plasma density at the center of the wafer support pedestal and strengthen it near the periphery for uniformity. Chosen to achieve.

  [69] Such cusp-type magnetic field tunability is achieved by providing at least a second overhead coil of a different diameter (eg, smaller) than the first coil in accordance with our findings. The The DC current of each coil is highly flexible and can be independently adjusted so that the shape of the cusp-type magnetic field can be brought closer to the desired plasma ion distribution with virtually any change in the surrounding plasma ion distribution It is. This choice of magnetic field shape can be designed to modify the central high or central low plasma ion concentration distribution.

  [70] One of the possible advantages is that (as mentioned above) the cusp-type magnetic field has a large radial gradient with respect to the magnetic field strength, thus doubling the correction pressure on the plasma with high efficiency. is there. However, since the magnetic field is constant over time, the tendency to generate arcing is much less so that a somewhat stronger magnetic field can be used for fairly large correction capacities when needed. As will be discussed later in this specification, this feature can be greatly useful at higher chamber pressures.

  [71] FIG. 1A illustrates a capacitively coupled plasma reactor that enables provision of an adjustable cusp-type magnetic field. The reactor of FIG. 1A includes a cylindrical side wall 5, a sealing 10 that is a gas distribution plate, and a wafer support pedestal 15 that holds a semiconductor workpiece 20. The ceiling 10 or gas distribution plate can be conductive or have an anode attached thereto so that it can act as an anode. The sealing 10 or gas distribution plate is typically made of aluminum and has an inner gas manifold and a gas injection orifice on its inner surface facing into the chamber. A process gas supply 25 provides process gas to the gas distribution plate 10. A vacuum pump 30 controls the pressure inside the reactor chamber. An RF generator in which plasma source power for igniting and maintaining the plasma inside the reactor chamber is connected to the wafer support pedestal 15 via an impedance matching circuit 45 so that the wafer support pedestal 15 acts as an RF electrode. 40. To act as a counter electrode, an anode (which can be a ceiling 10 made of a conductive material) is connected to RF ground. Such reactors tend to have a very non-uniform plasma ion concentration distribution, which is typically center high.

  [72] FIG. 1B shows that the ceiling 10 is connected through an RF impedance matching element 11 (shown only schematically) to a VHF signal generator 12 that provides plasma source power, rather than being directly connected to ground as in FIG. 1A. The characteristic of being is illustrated. In this case, the RF generator 40 simply controls the RF bias on the semiconductor wafer or workpiece 20. (The RF impedance matching element 11 can be a fixed tuning element such as a coaxial tuning stub or stripline circuit, for example.) Such features are discussed in more detail later in this specification.

  [73] A set of induction coils is provided above the ceiling to control the distribution of plasma ion concentration. In the case of FIG. 1A, this set of coils includes an inner coil 60 and an outer coil 65 that are coaxial with the cylindrical chamber and each constitute one winding of a conductor. Although the windings 60 and 65 are illustrated as one winding in FIG. 1A, for example, as shown in FIG. 1B, the windings 60 and 65 may include a plurality of windings arranged in the vertical direction. Alternatively, as shown in FIG. 1C, the windings 60, 65 may extend in both the vertical and horizontal directions. In the case of FIG. 1A, the inner coil 60 is disposed far above the ceiling 10 than the outer coil 65. However, in other cases, this arrangement can be reversed, or the two coils 60, 65 can be at the same height above the ceiling 10.

  [74] In the case of FIGS. 1A and 1B, the controller 90 controls the respective independent DC current supply 70, 75 connected to each of the coils 60, 65, thereby causing the overhead coils 60, 65 to respectively Determine the magnitude and polarity of the flowing current. Referring to FIG. 2, the controller 90 manages the DC current from the DC current supply 76 that supplies current through the controller 90 to the coils 60 and 65, and the controller 90 is connected to each of the coils 60 and 65. It is shown in the figure. In either case, the controller 90 can cause DC currents of different polarity and magnitude to flow through different ones of the coils 60,65. In the case of FIG. 2, the controller 90 adjusts the polarity of the DC current applied to each of the coils 60 and 65 and the pair of potential difference types 82 a and 82 b that adjust the DC current applied to the coils 60 and 65. And a pair of interlocking switches 84a and 84b that are determined independently. A programmable device, such as a microprocessor 91, can be included in the controller 90 to intelligently manage the potentiometers 82a, 82b and the interlock switches 84a, 84b.

  [75] The arrangement of the two coils 60, 65 illustrated in FIGS. 1A, 1B and 1C has certain advantages in that the inner coil 60 is placed higher above the ceiling 10 than the outer coil 65. I will provide a. In particular, the radial component of the magnetic field gradient provided by any coil is at least roughly proportional to the radius of the coil and inversely proportional to the axial displacement from the coil. Thus, the inner and outer coils 60, 65 play different roles due to their different sizes and displacements. The outer coil 65 dominates the entire surface of the wafer 20 due to its large radius and proximity to the wafer 20, while the inner coil 60 exerts its maximum effect near the center of the wafer, and is a trim coil for fine adjustment or shaping of the magnetic field. Can be considered. Other arrangements are possible to achieve such differential control with different coils placed at different radii with displacements from different plasmas. As will be described later with reference to the example embodiment herein, different changes to the ambient plasma ion concentration distribution are not only selecting different magnitudes of current through the respective overhead coils (60, 65), but also different. It can also be obtained by selecting a different polarity or direction of the overhead coil current.

  [76] FIG. 3A illustrates the radial (solid line) and azimuth (dotted line) components of the magnetic field generated by the inner coil 60 as a function of radial position on the wafer 20 in the case of FIG. 1A. FIG. 3B illustrates the radial (solid line) and azimuth (dotted line) components of the magnetic field generated by the outer coil 65 as a function of radial position on the wafer 20. The data illustrated in FIGS. 3A and 3B shows that the wafer 20 is 300 mm in diameter, the inner coil 60 is 12 inches in diameter and is about 10 inches above the plasma, and the outer coil 65 is 22 inches in diameter and about 6 inches above the plasma. Obtained in an implementation placed in FIG. 3C is a schematic illustration of a half-cusp magnetic field line pattern generated by the inner and outer overhead coils 60,65.

  [77] The controller 90 of FIG. 2 can vary the current applied to the respective coils 60, 65 to adjust the magnetic field at the wafer surface and thereby change the spatial distribution of plasma ion concentration. It is. Illustrated there are different coils 60, 65 to illustrate how the controller 90 can change these magnetic fields to affect and improve the plasma ion distribution in the chamber. It is the effect of different magnetic fields applied by one. In the following examples, the etch rate spatial distribution across the wafer surface is measured directly rather than the plasma ion distribution. The etch rate distribution changes directly with changes in the plasma ion distribution, so changes in one reflect changes in the other.

  [78] FIGS. 4A, 4B, 4C and 4D illustrate the beneficial effects achieved using the inner coil 60 only at a low chamber pressure (30 mT). FIG. 4A illustrates the etch rate (vertical Z axis) measured as a function of position on the surface of the wafer 20 (horizontal X and Y axes). Thus, FIG. 4A illustrates the spatial distribution of etch rate in the plane of the wafer surface. The central high non-uniformity of the etch rate distribution is clearly seen in FIG. 4A. FIG. 4A corresponds to the case where no magnetic field is applied, and therefore illustrates a non-uniform etch rate distribution that is unique to the reactor and needs to be corrected. The etching rate in this case has a standard deviation of 5.7%. 4 and 5, it is natural that the radial magnetic field is effective for improving the uniformity of the radial distribution of the plasma ion concentration. However, the magnetic field strength is described as an axial magnetic field near the center of the wafer. The axial magnetic field was chosen in this description because it is more easily measured. The radial magnetic field at the edge of the wafer is typically about 1/3 of the axial magnetic field at this location.

  [79] FIG. 4B illustrates how the etch rate distribution changes when the inner coil 60 is energized to generate a 9 Gauss magnetic field. The non-uniformity is reduced to a standard deviation of 4.7%.

  [80] In FIG. 4C, it can be seen that the magnetic field of the inner coil 60 has been increased to 18 Gauss, and the central peak has been greatly reduced, resulting in a standard deviation in etch rate across the wafer of 2.1%. Was reduced to.

  [81] In FIG. 4D, the magnetic field of the inner coil 60 has been further increased to 27 Gauss, and the central high pattern of FIG. 4A has been substantially inverted to the central low pattern. The standard deviation of the etching rate across the wafer surface in the case of FIG. 4D was 5.0%.

  [82] FIGS. 5A, 5B, 5C and 5D illustrate the beneficial effect of using both coils 60, 65 at higher chamber pressure (200 mT). FIG. 5A corresponds to FIG. 4A and depicts the central high etch rate non-uniformity of the reactor that is not corrected by the magnetic field. In this case, the standard deviation of the etching rate across the wafer surface was 5.2%.

  [83] In FIG. 5B, the outer coil 65 is energized to generate a 22 Gauss magnetic field, with some reduction in the central peak in the etch rate distribution. In this case, the standard deviation of the etching rate is reduced to 3.5%.

  [84] In FIG. 5C, both coils 60, 65 are energized to generate a 24 Gauss magnetic field. The result seen in FIG. 5C was that the central peak in the etching rate distribution was greatly reduced while the etching rate near the periphery increased. The overall effect is a more uniform etch rate distribution with a low standard deviation of 3.2%.

  [85] In FIG. 5D, both coils 60, 65 are energized to generate a 40 Gauss magnetic field, resulting in overcorrection, thus transforming the etch rate distribution across the wafer surface into a central low pattern. In this latter case, the standard deviation of the etching rate is slightly increased to 3.5% (compared to the case of FIG. 5C).

  [86] Comparing the results obtained in the low pressure test of FIGS. 4A-4D with the high pressure test of FIGS. 5A-5D, higher chamber pressures are required to achieve a similar correction for the non-uniform distribution of etch rates. It appears to require a larger magnetic field. For example, at 30 mT, the optimal correction was obtained using only the 18 Gauss inner coil 60, while at 300 mT, 24 Gauss using both coils 60, 65 to achieve optimal correction. A magnetic field is required.

  [87] FIG. 6 shows that the magnetic field of the overhead coil has a large effect on the plasma ion concentration or the uniformity of the etch rate distribution, but does not significantly affect the etch rate itself. This is an advantage because it is desirable to improve the uniformity of the etch rate distribution, but not to change the etch rate chosen for a particular semiconductor process. In FIG. 6, the etching rate at which the diamond symbol was measured (left vertical axis) is plotted as a function of the magnetic field (horizontal axis), and the square symbol is the standard deviation (non-uniformity) of the etching rate (right vertical axis). As a function of the magnetic field. The non-uniformity change in the range shown is about an order of magnitude and the change in etch rate is only about 25%.

  [88] The overhead coil inductors 60, 65 of FIGS. 1A, 1B, and 1C can be used with a conventional MERIE reactor. FIGS. 7 and 8 illustrate the case corresponding to FIG. 1A with the additional features of four conventional MERIE electromagnets 92, 94, 96, 98 and MERIE current controller 99. FIG. A current controller 99 provides AC current to each MERIE electromagnet 92, 94, 96, 98. Although each current has the same low frequency, their phase is offset by 90 degrees to generate a slowly rotating magnetic field in the chamber in a conventional manner.

Controlling plasma distribution using an overhead coil:
[89] According to the method of the reactor, the plasma ion concentration distribution across the wafer surface inherent to a particular reactor is tailored by selecting a particular magnetic field generated by the overhead coils 60, 65 in a particular manner. For example, the plasma distribution can be tailored to produce a more uniform etch rate distribution across the wafer surface. This alignment is accomplished, for example, by programming the controller 90 to select the optimal polarity and amplitude of the DC current in the overhead coil. Although this example relates to a reactor with only two concentric overhead coils (ie, coils 60, 65), this method can be performed with more than two coils, and the more overhead coils, the more accurate Results are obtained. The magnetic field is adjusted by the controller 90 changing the plasma ion concentration distribution across the wafer surface, which affects the etch rate distribution.

  [90] The first step measures the etch rate distribution from the overhead coils 60, 65 to the wafer surface without a correction magnetic field. The next step determines the change in plasma ion concentration distribution that makes the etch rate distribution more uniform. The last step determines a magnetic field that produces the desired change in plasma ion concentration distribution. Considering this magnetic field, the magnitude and direction of the current in the overhead coils 60, 65 required to generate such a magnetic field can be calculated from known static magnetic field equations.

  [91] The present inventors have sought a method for calculating the pressure acting on the plasma by the magnetic field of the overhead coils 60 and 65 (so-called “magnetic pressure”) from the magnetic field. This is discussed below. The magnetic pressure of the plasma changes the plasma ion concentration distribution. This change in the plasma ion concentration distribution proportionally changes the etching rate distribution over the wafer surface and can be observed directly. Thus, the plasma ion concentration distribution and the etch rate distribution across the wafer surface are related at least roughly by a proportionality factor.

  [92] First, the spatial distribution of the etch rate across the wafer surface is measured before the magnetic field is applied from the overhead coils 60,65. From this, it is possible to determine the desired change (to achieve a uniform distribution) of the etch rate distribution. Next, as a function of the position in the chamber and the current in the coil, the spatial distribution of the magnetic field generated by each overhead coil 60, 65 is analytically determined from the arrangement of each coil. By applying a known current set to the coil and measuring the resulting change in etch rate distribution across the wafer surface, the linear scale factor etches the vector sum of the magnetic fields of all coils on the wafer surface into the wafer surface. It can be estimated to relate to changes in velocity distribution. (This scale factor is generally a function of the plasma neutral pressure and is effective up to a chamber pressure of about 500 mT.) Thus, the desired change or correction (to achieve better uniformity) of the etch rate distribution is achieved. When considered, the required magnetic field can be determined (with the technique described later in this specification), and the corresponding coil current can be inferred therefrom using the analytically determined magnetic field spatial distribution function.

  [93] Desired corrections for etch rate distribution non-uniformities can be established in various ways. For example, a two-dimensional etch rate distribution across the wafer surface can be subtracted from the uniform or average etch rate to generate a “difference” distribution. The etch rate distribution non-uniformity to be corrected in this method is due to various factors within the reactor chamber, including non-uniform application of capacitively coupled source power, non-uniform process gas distribution, and non-uniform plasma ion concentration distribution. Is the result of In the above method, the non-uniformity is corrected by changing the plasma ion concentration distribution by the magnetic pressure.

  [94] The following method can also be used to establish a non-uniform “corrected” plasma distribution in some desired manner. In this case, the correction to be made is “uncorrected” or the difference between the ambient plasma ion concentration distribution and the desired distribution (which is itself non-uniform). Thus, this method helps to make the plasma density distribution more uniform or to a specific selected density distribution pattern that is not necessarily uniform.

  [95] A series of steps for performing the above method will be described with reference to FIG.

[96] The first step (block 910 of FIG. 9) analytically determines, for each of the overhead coils 60, 65, the wafer surface magnetic field equation as a function of the current in the coil and the radial position on the wafer surface. decide. Using cylindrical coordinates, this equation can be written with the i-th coil as B _i (r, z = wafer, I _i ). It is determined directly from Biosavart's law.

  [97] The next step (block 920 in FIG. 9) is performed without passing current through the overhead coils 60,65. In this step, the spatial distribution of plasma ion concentration across the wafer surface is measured. This spatial distribution can be written as n (r, z = wafer). In this step, the plasma ion concentration distribution can be indirectly measured by measuring the etch rate distribution across the wafer surface of the test wafer. A skilled technician can easily estimate the plasma ion concentration distribution from the etching rate distribution.

[98] Next, at block 930, a correction c (r) is determined for the measured plasma ion concentration spatial distribution function n (r, z = wafer) measured in the previous step. The correction c (r) can be defined in any suitable manner. For example, it can also be defined as the maximum value n (r, z = wafer) _max −n (r, z = wafer). In this way, c (r) is added to n (r, z = wafer) to generate a “corrected” distribution with a uniform amplitude equal to n (r) _max . Of course, the correction function c (r) can be defined differently to produce different uniform amplitudes. Or, as briefly mentioned above, if the desired distribution is non-uniform, the correction is the difference between the desired distribution and n (r, z = wafer).

[99] The next step (block 940) selects a “test” current I _i for each of the overhead coils 60, 65 and applies that current to the appropriate coil to measure the resulting plasma ion distribution. It can be written as n (r, z = wafer) _test . The change Δn (r) of the ion distribution can be approximated by subtracting the ion distribution measured with and without a magnetic field.

Δn (r) = n (r, z = wafer) −n (r, z = wafer) _test

[100] The next step (block 950) calculates a scale factor S that relates the pressure gradient exerted by the magnetic field (ie, magnetic pressure) to the change in ion distribution Δn (r). This calculation is done by dividing the magnetic pressure gradient by Δn (r). The magnetic pressure gradient of the i-th coil magnetic field B (r, z = wafer, I _i ) is calculated independently for each of the coils according to the magnetohydrodynamic equation.

▽ _r P =-▽ _r [B (r, z = wafer, I _i ) ² / 2μ ₀ ]

Here, the subscript r means a radial component. Thus, the results obtained independently for each coil are added together. Therefore, the total magnetic pressure gradient is

− ▽ _r {Σ _i [B (r, z = wafer, I _i ) ² / 2μ ₀ ]}

Therefore, the scale factor S is

S = {− ▽ _r {Σ _i [B (r, z = wafer, I _i ) ² / 2μ ₀ ]}} / Δn (r)

[101] This division operation may be performed with different values of r, and the results are averaged to obtain S in scalar form. Otherwise, the scale factor S is a function of r and is used appropriately.

[102] The scale factor S determined in the step of block 950 is the relationship between the coil current I _i that determines the magnetic pressure and the resulting change in ion distribution. Specifically, given a set of coil currents I _i , the corresponding ion distribution change n (r) can be calculated by multiplying the magnetic pressure determined from the set of I _i by a scale factor S.

Δn (r) = {− ▽ _r {Σ _i [B (r, z = wafer, I _i ) ² / 2μ ₀ ]}} / S

[103] This is the set of coil currents I _i at which a computer (such as the macro processor 91) uses the above equation to generate the best approximation of the change Δn (r) of the plasma ion concentration distribution previously defined or desired. Provides the basis for the next step (block 960) to search for. In this case, the desired change is equal to the correction function c (r) calculated in the block 930 step. In other words, the computer searches for a set of coil currents I _{i that} satisfy the following conditions:

{− ▽ _r {Σ _i [B (r, z = wafer, I _i ) ² / 2μ ₀ ]}} = c (r) s

For example, this search can be performed with known optimization techniques including steepest descent. Such techniques are easily performed by skilled artisans and need not be described here.

[104] The magnitude and polarity of the set of coil currents I _i discovered by the search are then sent to the controller 90 to apply these currents to the respective coils 60,65.

  [105] FIG. 10 compares magnetic pressure (solid line) with changes in plasma ion distribution (dotted line) measured as a function of radial position on the wafer surface. As described above, the magnetic pressure is the square gradient of the overhead coil magnetic field. FIG. 10 shows that there is a good correlation between changes in magnetic pressure and ion concentration distribution.

  [106] The application of such a method is illustrated in FIGS. FIG. 11 illustrates how the non-uniformity or standard deviation (vertical axis) in the etch rate space distribution on the wafer surface varied with the coil current of one of the overhead coils. At zero coil current, the standard deviation was about 12% and the ion distribution was center high as shown in FIG.

  [107] A minimum non-uniformity of about 3% was achieved with a coil current of about 17 amps. This represents an improvement of about 4 times (ie 12% to 3% standard deviation in the etch rate distribution). The actual or measured etch rate distribution is shown in FIG. 13A, and the etch rate distribution predicted using the technique of FIG. 9 is shown in FIG. 13B.

  [108] At a high coil current of 35 amps, the standard deviation of the etch rate distribution was about 14%. The measured etch rate space distribution is shown in FIG. 14A and the predicted distribution is shown in FIG. 14B.

  [109] Referring again to FIG. 13A, the most uniform ion distribution obtained is certainly not flat, and is actually a “bowl” shape with a concave near the periphery and a convex near the center. With more independent overhead coils (eg, 3 or more), it may be possible to perform current optimization that results in higher resolution and better uniformity. Thus, the reactor is not limited to having only two coils. The reactor can be implemented with various results using less than two or more than three overhead coils.

[110] The same method can be applied to control the plasma ion concentration distribution or etch rate distribution at the sealing surface. Such an approach is useful, for example, during the chamber cleaning process. FIG. 15 illustrates a version of the method of FIG. 9 with optimized ion concentration spatial distribution (or etch rate). The steps of FIG. 15, ie, blocks 910 ′, 920 ′, 930 ′, 940 ′, 950 ′, and 960 ′, are the steps of FIG. 9, except that they are performed for the sealing plane rather than the wafer plane. Same as 910, 920, 930, 940, 950 and 960. The first step (block 910 ′ of FIG. 15) analytically determines the sealing surface magnetic field equation for each of the overhead coils 60, 65 as a function of the current in the coil and the radial position on the wafer surface. . Using cylindrical coordinates, this equation can be written with the i-th coil as B _i (r, z = ceiling, I _i ). It is determined from a simple static magnetic field equation and is a function of some constant such as the coil current I _i and the radial position r on the sealing surface as well as the radius of the coil and the distance z = ceiling between the coil and the inner surface of the ceiling. .

  [111] The next step (block 920 'of FIG. 15) is performed without passing current through the overhead coils 60,65. In this step, the spatial distribution of plasma ion concentration across the sealing surface is measured. This spatial distribution can be written as n (r, z = ceiling). In this step, the plasma ion concentration distribution can be measured by conventional probes or other indirect techniques.

[112] Next, at block 930 ′, a correction c ′ (r) is determined for the measured plasma ion concentration spatial distribution function n (r, z = ceiling) measured in the previous step. (Prime notation 'is used here to distinguish the calculation of FIG. 15 from that of FIG. 9 above and does not imply a derivative.) Define the correction c ′ (r) in any suitable manner. be able to. For example, it can be defined as the maximum value n (r, z = ceiling) _max -n (r, z = ceiling). In this way, c ′ (r) is added to n (r, z = ceiling) to generate a “corrected” distribution with a uniform amplitude equal to n (r) _max . Of course, the correction function c ′ (r) can be defined differently to produce different uniform amplitudes. Also, if a specific non-uniform distribution is desired, the correction is the difference between the uncorrected or ambient plasma distribution n (r, z = ceiling) and the desired non-uniform distribution. Thus, this method can be used to either establish a desired plasma ion concentration distribution having a particular non-uniform pattern or to establish a uniform plasma ion concentration distribution.

[113] The next step (block 940 ′) selects a “test” current I _i for each of the overhead coils 60, 65 and applies that current to the appropriate coil to measure the resulting plasma ion distribution. N (r, z = ceiling) _test can be written. The change Δn (r) in the ion distribution is obtained by subtracting the ion distribution measured with and without a magnetic field.

Δn ′ (r) = n (r, z = ceiling) −n (r, z = ceiling) _test

[114] The next step (block 950 ′) calculates a scale factor S ′ that relates the pressure gradient exerted by the magnetic field (ie, magnetic pressure) to the change in ion distribution Δn ′ (r). This calculation is done by dividing the magnetic pressure gradient by Δn ′ (r). The magnetic pressure gradient of the i-th coil's magnetic field B (r, z = ceiling, I _i ) is independently calculated for each of the coils according to the magnetohydrodynamic equation.

▽ _r P = − ▽ _r [B (r, z = ceiling, I _i ) ² / 2μ ₀ ]

Here, the subscript r means a radial component. Thus, the results obtained independently for each coil are added together. Therefore, the total magnetic pressure gradient is

− ▽ _r {Δ _i [B (r, z = wafer, I _i ) ² / 2μ ₀ ]}

Therefore, the scale factor S is

S ′ = {− ▽ _r {Σ _i [B (r, z = wafer, I _i ) ² / 2μ ₀ ]}} / Δn ′ (r)

[115] The scale factor S ′ determined in the step of block 950 ′ is the relationship between the coil current I _i that determines the magnetic pressure and the resulting change in ion distribution. Specifically, given a set of coil currents I _i , a corresponding ion distribution change n ′ (r) can be calculated by multiplying the magnetic pressure determined from the set of I _i by a scale factor S ′. is there.

Δn ′ (r) = {− ▽ _r {Σ _i [B (r, z = wafer, I _i ) ² / 2μ ₀ ]}} / S ′

[116] This is, (such as microprocessor 91) computer by using the equations, previously prescribed or desired plasma ion density distribution change [Delta] n 'of the coil current I _i for generating the best approximation (r) Provides the basis for the next step (block 960 ′) for searching the set. In this case, the desired change is equal to the correction function c ′ (r) calculated in the step of block 930 ′. In other words, the computer searches for a set of coil currents I _{i that} satisfy the following conditions:

{− ▽ _r {Σ _i [B (r, z = wafer, I _i ) ² / 2μ ₀ ]}} / c ′ (r) S ′

[117] This search can also be performed with known optimization techniques including, for example, steepest descent. Such techniques are easily performed by skilled artisans and need not be described here.

[118] The magnitude and polarity of the set of coil currents I _i discovered by the search are then sent to the controller 90 to apply these currents to the respective coils 60,65.

  [119] With only one overhead coil, it is possible to use this apparatus to optimize the uniformity of plasma ion distribution on either the wafer or the ceiling, not both simultaneously. With at least two overhead coils (eg, overhead coils 60, 65), it is possible to at least approximately optimize the plasma ion distribution uniformity simultaneously on both the wafer and the ceiling.

Plasma steering with overhead coils:
[120] The inventors have discovered that the coil current I _i can be selected by steering the plasma towards the ceiling and / or the sidewalls, or steering the wafer surface. Similar to the method of FIG. 9, the coil current I _i can also be selected to improve the uniformity of the plasma density distribution at the sealing surface. As a result, the plasma concentrates on the wafer during processing and concentrates on the sealing and / or sidewalls during cleaning. Since the plasma concentrates in this way, the cleaning time can be shortened.

  [121] In one example, the controller 90 applied a current of -17.5 amperes to the inner coil 60 and a current of +12.5 amperes to the outer coil 65 to steer the plasma to the chamber sidewall. FIG. 16 illustrates the radial portion within the chamber extending along a horizontal axis from a zero radius to the periphery of the chamber and extending along the vertical axis from the wafer surface to the ceiling. The small arrows in FIG. 16 indicate the chamber when the controller 90 applies a current of -17.5 amperes to the inner coil 60 and a current of +12.5 amperes to the outer coil 65 to steer the plasma to the sidewall of the chamber. Displays the magnitude and direction of the magnetic field at various positions. FIG. 17 illustrates the gradient of the corresponding magnetic field square of the wafer surface as a function of radial position.

  [122] In another example, the controller 90 applied a current of -12.5 amperes to the inner coil 60 and a current of +5 amperes to the outer coil 65 to steer the plasma to the roof of the chamber. FIG. 18 illustrates the radial portion within the chamber extending along a horizontal axis from zero radius to the periphery of the chamber and extending along the vertical axis from the wafer surface to the ceiling. The small arrows in FIG. 18 indicate that the controller 90 applies a current of −12.5 amperes to the inner coil 60 and a current of +5 amperes to the outer coil 65 to cause the plasma to be steered into the chamber sidewall. Displays the magnitude and direction of the magnetic field at various positions. FIG. 19 illustrates the corresponding magnetic field square gradient on the wafer surface as a function of radial position.

  [123] In another example, the controller 90 applies a current of -25 amperes to the inner coil 60 and a current of +2.75 amperes to the outer coil 65, causing the plasma to flow along the field line extending from the center of the ceiling to the sidewall. And steered. FIG. 20 illustrates the radial portion within the chamber extending along the horizontal axis from zero radius to the periphery of the chamber and extending along the vertical axis from the wafer surface to the ceiling. The small arrows in FIG. 20 indicate when the controller 90 applies a current of −25 amps to the inner coil 60 and a current of +2.2.5 amps to the outer coil 65 to steer the plasma to the chamber sidewall. Displays the magnitude and direction of the magnetic field at various locations within the chamber. FIG. 21 displays the gradient of the corresponding magnetic field square of the wafer surface as a function of radial position.

  [124] FIG. 17 shows that when the plasma is steered to the edge, a high positive magnetic pressure on the plasma is applied near the edge of the chamber. FIG. 19 shows that a low magnetic pressure is applied to the plasma near the edge of the chamber when the plasma is directed to the edge of the ceiling. FIG. 21 shows a high negative magnetic pressure near the chamber edge as the magnetic field lines extend from the ceiling to the edge.

  [125] In this manner, the current in the overhead coils 60, 65 can be selected to direct the plasma to various locations in the chamber that require cleaning of the ceiling, sidewalls, and the like. Alternatively, the plasma can be concentrated closer to the wafer. To steer the plasma to either the wafer or the ceiling, or to assign the plasma between the wafer and the ceiling according to a certain steering ratio SR, a method as illustrated in FIG. 22 can be performed.

[126] Referring to FIG. 22, the first step (2210 in FIG. 22) is an analytical analysis of the magnetic field inside the chamber as a function of all coil currents in the overhead coils (eg, the pair of coils 60, 65). Define the model. This is easily accomplished by skilled artisans using static magnetic field equations and need not be described here. The magnetic field is the sum of the independent magnetic fields from each coil. Each independent magnetic field is a function of the diameter of the respective coil, the position of each coil, the current in the coil, and the position in the chamber. Thus, the magnetic field generated by the i-th coil is written as follows:

B (x, y, z, I _i )

So the total magnetic field is

Σ _i {B (x, y, z, I _i )}

[127] The next step (block 2220) selects a set of magnetic fields that meet the desired set of process conditions. For example, to steer the plasma to the ceiling, the magnetic field is selected to generate a magnetic pressure on the plasma that pushes the plasma toward the ceiling, as illustrated in the example of FIG. To steer the plasma toward the sidewall, a magnetic field is selected to generate a magnetic pressure against the plasma that pushes the plasma toward the periphery, as illustrated in FIG.

  [128] For each magnetic field that satisfies the specific conditions defined in block 2220 above, the computer searches the model defined in block 2210 for a set of coil currents that generate the desired magnetic field. This is the next step after block 2230. Each current set determined in block 2230 is stored in a memory location associated with the corresponding process condition along with the name of the corresponding condition (block 2240 of FIG. 22). Whenever a specific process condition is selected (eg, steering the plasma to the ceiling), the microprocessor 91 retrieves a set of current values from the corresponding memory location (block 2250) and directs the corresponding current to the appropriate coil. Apply (block 2260).

  [129] FIG. 23 illustrates how the microprocessor 91 can be programmed to respond to user input. Initially, a determination is made as to whether the process includes etching of the wafer surface (block 2310 and whether the process includes sealing cleaning (etching) (block 2320)). If only the wafer is etched, the plasma is steered to the wafer (block 2330) and the uniformity of the plasma distribution on the wafer surface is optimized using the method of FIG. 9 (block 2350). If the wafer is etched at the same time that the ceiling is cleaned, a plasma density is assigned between the ceiling and the wafer (block 2360), and the plasma density uniformity is at the wafer surface as in FIG. Optimized by sealing as in block 15 (block 2370). If only the ceiling is cleaned, the plasma is steered to the ceiling (block 2380) and the plasma density uniformity at the ceiling is optimized (block 2390).

Use with VHF overhead electrode:
[130] FIG. 24 illustrates how the inner and outer coils 60, 65 can be combined with a capacitively coupled reactor having an overhead electrode connected to a VHF plasma source power generator through a fixed tuning stub. To do. Such a reactor is described in US Patent Application No. 10 / 028,922 “Plasma Reactor with Overhead RF Electric Tuned to the Plasma” filed December 19, 2001 by Daniel Hoffman et al. The disclosure of which is incorporated herein by reference.

  [131] Referring to FIG. 24, the plasma reactor includes a reactor chamber 100 with a wafer support 105 that supports a semiconductor wafer 110 at the bottom of the chamber. The process kit may include a conductive or semiconductive ring 115 supported by a dielectric ring 120 on a grounded chamber body 127 in an exemplary implementation. The chamber 100 is bounded at the top by a disk-shaped overhead conductor electrode 125 supported by a gap length above the wafer 110 on a chamber body 127 grounded by a dielectric seal. In one implementation, the wafer support 105 is movable in the vertical direction so that the gap length can be changed. In another implementation, the gap length is a fixed length. The overhead electrode 125 can be a metal (eg, aluminum) whose inner surface is covered with a metalloid material (eg, Si or SiC), or can itself be a metalloid. The RF generator 150 applies RF power to the electrode 125. RF power from generator 150 is coupled into a coaxial stub 135 connected to electrode 125 through a coaxial cable 162 aligned with generator 150. The stub 135 has a characteristic impedance, has a resonant frequency, and provides impedance matching between the electrode 125 and the output of the coaxial cable 162 or RF power generator 150, as described more fully below. The chamber body is connected to the RF return (RF ground) of the RF generator 150. The RF path from the overhead electrode 125 to the RF ground is affected by the capacitance of the dielectric seal 120 and the dielectric seal 130. Wafer support 105, wafer 110, and conductive or semi-conductive ring 115 of the process kit provide the main RF return for RF power applied to electrode 125.

  [132] As in the case of FIG. 1A, the inner coil 60 is on a plane that is smaller than half the diameter of the outer coil 65 and farther away from the chamber than the outer coil 65. The inner coil 60 is positioned significantly above the electrode 125, while the outer coil 65 is positioned on or near the top plane of the electrode 125. As in the case of FIG. 1, the DC current in the coils 60 and 65 is controlled by a plasma steering controller 90 that manages the current supply sources 70 and 75 of the coils 60 and 65.

  [133] The capacitance of the overhead electrode assembly 126, including electrodes 125, process kits 115, 120, and dielectric seal 130, measured with respect to RF retrace or ground, is 180 picofarads in the illustrative case. Electrode assembly capacity is affected by electrode area, gap length (distance between wafer support and overhead electrode), and by factors that affect stray capacitance, particularly the dielectric values of seal 130 and dielectric ring 120, It is affected by the dielectric constant and thickness of the material used. More generally, as discussed below, the capacitance of the electrode assembly 126 (an unsigned number or scalar) is the negative capacitance of the plasma (in magnitude at a specific source power frequency, plasma density, and working pressure). Complex number).

[134] Many of the factors that affect the above relationships are largely dictated by the plasma process requirements that need to be performed by the reactor, the size of the wafer, and the current state of the requirement that the process be performed uniformly on the wafer. Has been. Thus, while the plasma capacity is a function of the plasma density and the source power frequency, the electrode capacity is a function of the wafer support-electrode gap (height), the electrode diameter, and the dielectric value of the assembly insulator. . The plasma density, working pressure, gap, and electrode diameter must meet the requirements of the plasma process performed by the reactor. In particular, the ion concentration must be within a certain range. For example, silicon and dielectric plasma etch processes generally require plasma ion concentrations that are in the range of 10 ⁹ to 10 ¹² ions / cc. The wafer electrode gap provides optimal plasma ion distribution uniformity for an 8 inch wafer, for example, when the gap is about 2 inches. The electrode diameter is preferably at least about the same, if not exceeding the diameter of the wafer. Similarly, the working pressure has a practical range of typical etching and other plasma processes.

  [135] However, it has been found that there are still other factors that can be selected to achieve the preferred relationship described above. In particular, the selection of the source frequency and the capacity for the overhead assembly 126. If the source power frequency is selected to be the VHF frequency and the dielectric values of the insulating components of the electrode assembly 126 are properly selected, the above dimensional constraints imposed on the electrodes and the constraints imposed on the plasma (eg, density range). ), The electrode capacity can be matched with the negative capacity of the plasma. Such a selection can achieve a match or near match between the source power frequency and the plasma-electrode resonance frequency.

  [136] Thus, in the exemplary case, for an 8-inch wafer, the overhead electrode diameter is approximately 11 inches, the gap is approximately 2 inches, the plasma density and working pressure are typical values for the etching process described above, and VHF The source frequency is 210 MHz (although equally effective at other VHF frequencies), and the source power frequency, plasma electrode resonance frequency, and stub resonance frequency are all matched or nearly matched.

  [137] More specifically, to achieve a detuning effect that advantageously reduces system Q, these three frequencies are 210 MHz source power frequency, about 200 MHz electrode-plasma electrode resonance frequency, and stub resonance. The frequency is slightly offset from each other by about 220 MHz. Such a reduction in system Q makes the reactor performance independent of changes in conditions within the chamber and can be performed over a wider process window as the overall process becomes more stable.

  [138] Currently preferred modes include a 12 inch diameter wafer, a wafer-sealing gap of about 1.25 inches, and a chamber diameter suitable to accommodate a VHF source power frequency of 162 MHz (rather than 210 MHz above) Has a diameter.

  [139] The coaxial stub 135 is a specially designed design that further contributes to overall system stability, its wide process window capability and many other beneficial benefits. It includes an inner cylindrical conductor 140 and an outer concentric cylindrical conductor 145. The insulator 147 (indicated by cross hatching in FIG. 24) has a relative dielectric constant of 1, for example, and fills the space between the inner and outer conductors 140 and 145. Inner and outer conductors 140, 145 can be formed of, for example, nickel-coated aluminum. In the exemplary case, outer conductor 145 has a diameter of about 4 inches and inner conductor 140 has a diameter of about 1.5 inches. The stub characteristic impedance is determined by the radii of the inner and outer conductors 140 and 145 and the dielectric constant of the insulator 147. The stub 135 in this case described above has a characteristic impedance of 65. More generally, the stub characteristic impedance exceeds the source power output impedance by about 20% to 40%, preferably about 30%. The stub 135 has an axial length of approximately 29 inches (half wavelength at 220 MHz) so as to have a resonance near 220 MHz that is approximately matched with a slight offset from the VHF source power frequency of 210 MHz.

  [140] A tap 160 is provided at a particular point along the axial length of the stub 135 to apply RF power from the RF generator 150 to the stub 135, as discussed below. The RF power terminal 150b and the RF return terminal 150a of the generator 150 are connected to the inner and outer coaxial stub conductors 140, 145 by taps 160 on the stub 135, respectively. These connections are made through a generator-stub coaxial cable 162 having a characteristic impedance that matches the output impedance of the generator 150 (typically 50) in a known manner. A termination conductor 165 at the far end 135a of the stub 135 shorts the inner and outer conductors 140, 145 together, and the stub 135 is shorted at its far end 135a. At the proximal end 135b (non-shorted end) of the stub 135, the outer conductor 145 is connected to the chamber body via an annular conductive housing or support 175, and the inner conductor 140 is connected to the electrode 125 via a conductive cylinder or support 176. Connected to the center of the. A dielectric ring 180 is held between the cylinder 176 and the electrode 125 to separate them.

  [141] Inner conductor 140 provides conduits for utilities such as process gases and coolants. The main advantage of this feature is that, unlike typical plasma reactors, the gas line 170 and the coolant line 173 do not exceed a large potential difference. Thus, they can be constructed of metals that are cheaper and more reliable materials for such purposes. The metal gas line 170 supplies a gas outlet 172 in or adjacent to the overhead electrode 125, and the metal coolant line 173 supplies a coolant passage or jacket 174 in the overhead electrode 125.

  [142] This allows active and resonant impedance transformations to stub match this special configuration between the RF generator 150 and the overhead electrode assembly 126, plasma load handling, reflected power minimization, and wide variations in load impedance. Provided by a very wide impedance matching space. As a result, a wide process window and process flexibility is provided along with power utilization efficiency previously unobtainable, while minimizing or avoiding the need for typical impedance matching devices. As mentioned above, the stub resonance frequency is also offset from the ideal match in order to further enhance the overall system Q, system stability and process window and multi-process capability.

Matching of electrode-plasma resonance frequency and VHF source power frequency:
[143] As outlined above, the main features are the overhead electrode assembly 126 for resonating with the plasma at the electrode-plasma resonance frequency and for matching (or nearly matching) the source power frequency with the electrode-plasma frequency. Is to configure. The electrode assembly 126 has a significant capacitive reactance, which is a complex function of frequency, plasma density, and other parameters. (As will be described in more detail below, the plasma is analyzed for complex function reactance including an imaginary term and generally corresponds to a negative capacitance.) The electrode-plasma resonance frequency is determined by the electrode assembly 126 and the reactance of the plasma. (Similar to the resonance frequency of the capacitor / inductor resonance circuit determined by the reactance of the capacitor and the inductor). As described above, the electrode-plasma resonance frequency does not necessarily have to be the source power frequency so as to depend on the plasma density. Thus, the problem is that the source power such that the electrode-plasma resonance frequency is equal to or nearly equal to the source power frequency, given the constraints that the plasma reactance is actually limited to a specific range of plasma density and electrode dimensions. Finding the frequency. The problem is even more difficult because the plasma density (which affects plasma reactance) and electrode dimensions (which affect electrode capacity) must satisfy certain process constraints. Specifically, for dielectric and conductor plasma etching processes, the plasma density should be in the range of 10 ⁹ to 10 ¹² ions / cc, which is a plasma reactance constraint. Further, for example, for processing an 8 inch diameter wafer, a more uniform plasma ion concentration distribution is achieved with a wafer-electrode gap or height of about 2 inches and an electrode diameter on the order of or larger than the wafer diameter, which is the electrode capacity. It becomes a restriction. On the other hand, different gaps can be utilized for 12 inch diameter wafers.

[144] Thus, by matching (or nearly matching) the electrode capacitance to the magnitude of the negative capacitance of the plasma, the electrode-plasma resonance frequency and the source power frequency are at least substantially matched. For the general conductor and dielectric etch process conditions listed above (ie, plasma density of 10 ⁹ to 10 ¹² ions / cc, 2 inch gap and approximately 11 inch electrode diameter), the source power frequency is VHF. Matching is possible at any frequency. Other conditions (eg, different wafer diameters, different plasma densities, etc.) can dictate different frequency ranges to achieve such matching in the implementation of this feature of the reactor. As described in detail below, the plasma density described above under convenient plasma processing conditions for processing 8 inch wafers in several main applications, including dielectric and metal plasma etching and chemical vapor deposition. One exemplary embodiment with a plasma capacity was -50 to -400 picofarads. In an exemplary case, a dielectric material for seal 130 having a dielectric constant of 9 and a thickness of about 1 inch, using an electrode diameter of 11 inches, a gap length of about 2 inches (distance from the electrode to the table), and By choosing a dielectric material for the ring 120 having a dielectric constant of 4 and a thickness on the order of 10 mm, the capacity of the overhead electrode assembly 126 was matched to the magnitude of this negative plasma capacity.

  [145] If the capacitive matching is as described above, the electrode assembly 126 and plasma combination resonates at an electrode-plasma resonant frequency that is at least approximately matched to the source power frequency applied to the electrode 125. We can match or nearly match this electrode-plasma resonant frequency and source power frequency at VHF frequencies for convenient etch plasma processing methods, environments and plasmas, and such frequency matching or near matching. It has been found that it is very advantageous to implement. In an exemplary case, as described in detail below, the electrode-plasma resonance frequency corresponding to the aforementioned plasma negative capacity is approximately 200 MHz. Because the source power frequency is 210 MHz and the other advantages discussed below are realized, the source power frequency is offset slightly above the electrode-plasma resonance frequency and is approximately matched.

[146] Plasma capacity is a function of plasma electron density above all. This is related to plasma ion concentration and needs to be kept in the range of 10 ⁹ to 10 ¹² ions / cc to provide good plasma processing conditions. Since this density determines the plasma negative capacity along with the source power frequency and other parameters, its selection is constrained by the need to optimize plasma processing conditions, as described in more detail below. The overhead electrode assembly capacity includes, for example, the gap length (distance between the electrode 125 and the wafer), the area of the electrode 125, the range of the dielectric loss tangent of the dielectric seal 130, and the dielectric seal 130 between the electrode 125 and the grounded chamber body 127. The choice of dielectric constant, the choice of dielectric constant for the process kit dielectric seal 130, and many physical factors such as the thickness of the dielectric seals 130 and 120 and the thickness and dielectric constant of the ring 180 are affected. This allows some adjustment of the electrode assembly capacity through selections made among these and other physical factors that affect the overhead electrode assembly capacity. The inventors have discovered that this range of adjustment is sufficient to achieve the required degree of matching of the overhead electrode assembly capacity to the magnitude of the negative plasma capacity. In particular, the dielectric material and dimensions of seal 130 and ring 120 are selected to provide the desired dielectric constant and dielectric value results. Second, the same physical factors that affect electrode capacity, especially gap length, have the following practicality: the need to deal with large diameter wafers; doing so with good uniformity of plasma ion concentration distribution over the entire diameter of the wafer Despite the fact that it is determined or limited by having good control of the ion concentration versus ion energy, a matching of electrode capacity and plasma capacity can be achieved.

  [147] Considering the aforementioned plasma capacity range and overhead electrode capacity matching, the electrode-plasma resonance frequency was 200 MHz for a source power frequency of 210 MHz.

  [148] The great advantage of choosing the capacitance of the electrode assembly 126 in this way and then matching the resulting electrode-plasma resonance frequency to the source power frequency is that the resonance of the electrode and plasma near the source power frequency is Provides wider impedance matching and wider process window, resulting in greater resistance to changes in process conditions, and greater performance stability. The overall processing system is less sensitive to variations in operating conditions, such as plasma impedance shifts, and thus becomes more reliable as the range of process applicability increases. As will be discussed later in this specification, this advantage is further enhanced by a small offset between the electrode-plasma resonance frequency and the source power frequency.

  [149] FIG. 25 shows a capacitively coupled reactor having an overhead electrode connected to a VHF plasma source power generator through fixed tuning stubs with internal and external coils 60, 65 and having a MERIE electromagnet around it. Illustrates how they can be combined. Such a reactor is described in US Patent Application No. 10 / 028,922 “Plasma Reactor with Overhead RF Electric Tuned to the Plasma” filed December 19, 2001 by Daniel Hoffman et al. The disclosure of which is incorporated herein by reference.

  [150] Referring to FIG. 25, the VHF capacitively coupled plasma reactor includes the following elements found in FIG. 1A. The reactor chamber 100 includes a wafer support 105 that supports a semiconductor wafer 110 at the bottom of the chamber. The illustrated process kit consists of a conductive or semiconductive ring 115 supported by a dielectric ring 120 on a grounded chamber body 127. The chamber 100 is bounded at the top by a disk-like overhead aluminum electrode 125 supported at a predetermined gap length above the wafer 110 on the chamber body 127 grounded by a dielectric seal 130. The overhead electrode 125 can also be a metal (eg, aluminum) whose inner surface is covered with a metalloid material (eg, Si or SiC), or can itself be a metalloid material. The RF generator 150 applies RF power to the electrode 125. RF power from the generator 150 is coupled into a coaxial stub 135 to which the electrode 125 is connected through a coaxial cable 162 aligned with the generator 150. The stub 135 has a characteristic impedance, a resonant frequency, and provides impedance matching between the electrode 125 and the coaxial cable 162 / RF power generator 150, as described more fully below. The chamber body is connected to the RF return (RF ground) of the RF generator 150. The RF path from the overhead electrode 125 to the RF ground is affected by the capacitance of the dielectric ring 120 and dielectric seal 130 of the process kit. Wafer support 105, wafer 110, and conductive (or semiconductive) ring 115 of the process kit provide the main RF return for RF power applied to electrode 125.

  [151] As in FIG. 1A, the inner coil 60 is on a plane that is less than half the diameter of the outer coil 65 and farther from the chamber than the outer coil 65. The inner coil is positioned significantly above the electrode 125, while the outer coil 65 is positioned on or near the top plane of the electrode 125. As in the case of FIG. 1, the DC current in the coils 60 and 65 is controlled by a plasma steering controller 90 that manages the current supply sources 70 and 75 of the coils 60 and 65.

  [152] By introducing a set of equally spaced MERIE electromagnets 902 around the periphery of the wafer support pedestal and outside the reactor chamber (as shown in FIGS. 7 and 8), An improvement in uniformity is achieved. These MERIE magnets are generally adapted to generate a magnetic field that rotates slowly about the axis of symmetry of the cylindrical chamber over the surface of the wafer support pedestal. In one case, this feature is realized by a MERIE magnet 902 having electromagnet windings wound about respective axes tangent to the periphery of the wafer support pedestal. In this case, the MERIE current controller 904 controls the independent current of each MERIE magnet. The circulating magnetic field is processed by a controller 904 that provides independent AC current to each independent magnet winding at the same frequency but offset in phase by 90 degrees (or 360 degrees divided by the number of MERIE magnets). It is generated in the plane of the support. In another case, the rotating magnetic field feature is realized by a support frame 1020 (dotted line) that supports all MERIE magnets rotated about the axis of symmetry by the rotor 1025 (dotted line). In this other case, the MERIE magnet is a permanent magnet.

  [153] A second array of MERIE magnets 906 (shown in dotted lines) equally spaced around the workpiece or wafer support pedestal can also be provided in a higher plane than the first set of MERIE magnets 902. Both sets of magnets are in respective planes near the plane of the workpiece support.

  [154] The controller 910 applies a low frequency (0.5-10 Hz) AC current to each electromagnet 902,906. The phase of the current applied to adjacent magnets is offset by 90 degrees as described above. The result is that the magnetic field rotates around the axis of symmetry of the workpiece support at a low frequency of AC current. The magnetic field pulls the plasma toward the magnetic field near the workpiece surface and circulates in the magnetic field. This stirs the plasma and makes its density distribution more uniform. As a result, more uniform etching results are obtained over the entire surface of the wafer, greatly improving the performance of the reactor.

Combination overhead electrode and gas distribution plate:
[155] It is desirable to supply process gas from an overhead ceiling to improve the uniformity of gas distribution within the chamber. For this purpose, the overhead electrode 125 in the case of FIGS. 24 and 25 can be a gas distribution showerhead, and therefore a number of gas injection ports or small holes 300 on its bottom surface facing the workpiece support 105. Have In the exemplary case, the holes 300 are between 0.01 and 0.03 inches in diameter, and their centers are uniformly spaced by about 3/8 inch.

  [156] An overhead electrode / gas distribution plate 125 (hereinafter gas distribution plate 125) improves resistance to arcing. This is due to the introduction of arc suppression features that exclude process gases and / or plasma from the center of each opening or hole 300. The arc-suppressing feature is the center piece or disk at the center of the hole 300 supported at the end of each cylindrical finger or narrow rod 303 as shown in the cross-sectional view of FIG. 26 and the enlarged cross-sectional view of FIG. 302 sets. Arc discharge in a typical gas distribution plate occurs near the center of the gas injection hole. Therefore, by placing the center piece 302 in the center of each hole 300, the process gas is prevented from reaching the center of each hole 300, and the occurrence of arc discharge is reduced. As shown in the plan view of FIG. 28, by introducing the center piece 302 into the hole 300, otherwise the circular opening or hole 300 is transformed into an annular opening.

  [157] Referring to FIG. 29A, a gas distribution plate 125 with improved arc suppression comprises a cover 1402 and a base 1404. The base 1404 is formed with a disk-like plate 1406 having a gas injection opening surrounded by an annular wall 1408 having an inner shoulder 1410 passing through the inside. The cover 1402 is also a disk-shaped plate. The disc 302 is the end of a cylindrical finger 303 that is attached to the bottom surface of the cover 1402 and extends downwardly therefrom. The outer edge of the cover 1402 rests on the shoulder 1410 of the base 1404, forming a gas manifold 1414 (FIG. 26) between the cover 1402 and the base 1404. Process gas flows into the manifold 1414 from the gas inlet 1416 at the center of the cover 1402.

  [158] The portion of the gas distribution plate 125 that contacts the process gas or plasma in the chamber can be formed of a metal such as aluminum coated with a semiconductor processing compatible material such as silicon carbide. In this example, the entire surface of the gas distribution plate except the top surface of the cover 1402 is covered with a silicon carbide coating 1502 as shown in the enlarged partial cross-sectional view of FIG. 29B. As shown in FIG. 30, the aluminum top surface of the cover 1402 is in contact with a temperature control member 1520 that is water cooled by a water jacket with a coolant circulated by a heat exchanger 1524, and the heat of the gas distribution plate 125. The conductive aluminum material has a controlled temperature. Alternatively, the water jacket can be within the gas distribution plate 125 as shown in FIG.

  [159] However, in order for the silicon carbide coating 1502 to have the same controlled temperature, there must be a thermally conductive bond between the silicon carbide coating and the aluminum. Otherwise, the temperature of the silicon carbide coating may fluctuate uncontrollably. In order to achieve good heat conduction between the aluminum material of the gas distribution plate 125 and the silicon carbide coating, a polymer adhesive layer 1504 is provided between the aluminum gas distribution plate and the silicon carbide coating 1502 as shown in FIG. 29A. Is formed. FIG. 29A shows that the polymer adhesion layer 1504 is between the silicon carbide coating 1502 and the aluminum base 1404. The polymer adhesion layer provides good heat conduction between the aluminum and the silicon carbide coating 1502, and the temperature of the coating 1502 is controlled by the heat exchanger 1524.

  [160] FIGS. 32, 33, and 34 illustrate how the gas distribution plate 125 of FIG. 29A can be modified to provide dual zone gas flow control. Such a feature can be used to correct either the central high or central low etch rate or deposition rate spatial distribution by selecting a complementary process gas distribution. Specifically, an annular divider or wall 1602 divides the gas manifold 1414 into a central manifold 1414a and an outer manifold 1414b. In addition to the central gas supply 1416 supplied to the central manifold 1414a, another gas supply 1418 between the center and the periphery of the gas distribution plate 125 supplies the outer manifold 1414b. The dual zone controller 1610 distributes the gas flow from the process gas supply 1612 between the inner and outer gas supplies 1416, 1418. FIG. 35 illustrates one implementation of valve 1610 in which articulation vanes 1618 control the respective amount of gas flow to the inner and outer gas manifolds 1414a, 1414b of the gas distribution plate. Intelligent flow controller 1640 manages the position of vane 1618. In another implementation shown in FIG. 36, a pair of valves 1651, 1652 provides independent gas flow control for each radial zone of the chamber.

  [161] FIG. 37 illustrates the case where the gas distribution plate 125 has three gas flow zones. Manifold 1414 is separated into three manifolds 1414a, 1414b and 1414c by inner and outer annular partitions 1604, 1606. Three respective gas supplies 1416, 1418, 1420 provide gas flow to respective manifolds 1414a, b, c.

  [162] Although various cases have been described in this specification as having a pair of overhead coils 60, 65, FIG. 37 shows that more than two overhead coils are possible. Indeed, the case of FIG. 37 is illustrated as having three concentric overhead coils or coils 60, 64 and 65. By increasing the number of independently controlled overhead coils, it is felt that the processing non-uniformity is corrected and the resolution is increased.

  [163] The multi-zone gas distribution plate of FIGS. 34 and 37 enjoys the advantage of flexible control over gas distribution between the inner and outer processing zones of the workpiece. However, another way to customize the gas flow is to do so permanently by providing different gas injection hole sizes at different radii of the gas distribution plate 125. For example, if the reactor tends to present a center-high spatial etch rate distribution, by using a small gas injection hole 300 in the center and a large one near the periphery, there will be less gas near the center of the chamber and less at the periphery. More gas is supplied. Such a gas distribution plate is shown in the plan view of FIG. For the central low etch rate distribution, the reverse hole arrangement will be used as illustrated in FIG.

Plasma steering in the reactor of FIG. 9:
[164] Plasma steering as described above with reference to FIGS. 11-14 is performed in the case of FIG. A magnetic field pointing to the sidewall is generated by applying a -13 ampere current to the inner coil 60 and a +1.4 ampere current to the outer coil 65. A magnetic field toward the periphery of the ceiling or electrode 125 is generated by applying a -13 ampere current to the inner coil 60 and a +5.2 ampere current to the outer coil 65. A dense magnetic field at the sidewall is generated by applying a -13 ampere current to the inner coil 60 and a +9.2 ampere current to the outer coil 65. The inventors have found that the etching rate of the chamber surface during cleaning is improved by about 40% by applying a magnetic field directed to the periphery of the sealing or electrode 125 in the above-described manner.

Coil shape:
[165] Although the above case has been described with reference to inner and outer coils 60, 65, a greater number of coils can be used. For example, the case of FIG. 40 has five overhead coils 4060, 4062, 4064, 4066, and 4068, each having a respective current controlled separately by the controller 90. The coils 4060, 4062, 4064, 4066, 4068 can be the same height above the ceiling 125 (as in FIG. 40) or different heights. FIG. 41 illustrates the case where the overhead coils 60 and 65 are at the same height. In FIG. 41, the windings of the coils 60 and 65 are overlapped both in the vertical direction and in the radial direction. 42 and 43 illustrate different cases where the coils 60, 65 have windings extending in the vertical and radial directions.

  [166] As discussed herein above with reference to FIG. 1A, the magnetic pressure on the plasma to correct for the non-uniform distribution is proportional to the radial component of the gradient of the square of the magnetic field. Thus, the most efficient approach is to use a magnetic field with a large radial gradient, such as a cusp-type magnetic field. As further discussed above, the greater the efficiency of a cusp-type magnetic field, the less magnetic field strength is required for a given amount of magnetic pressure, thereby reducing device damage associated with high magnetic fields. Or disappear. FIG. 44 illustrates the case where a complete cusp-type magnetic field is generated by a pair of coils 4420 and 4440 disposed above and below the chamber, respectively. The currents in the top and bottom coils 4420, 4440 are clockwise and counterclockwise, respectively. FIG. 45 is a simplified illustration of a magnetic line pattern of a complete cusp-type magnetic field generated by a pair of coils 4420, 4440.

[167] FIG. 46 illustrates the case where four electromagnets 4610, 4620, 4630, 4640 of a conventional MERIE reactor 4650 are used to generate the full cusp-type magnetic field of FIG. A current controller 4660 that controls the current in each electromagnet 4610, 4620, 4630, 4640 has the same DC current (eg, clockwise) as indicated by the arrows in FIG. 46 in all electromagnets 4610, 4620, 4630, 4640. ) Programmed to flow in the direction. In this way, the DC current in the top conductors 4610a, 4620a, 4630a, 4640a forms a clockwise current loop, and the DC current in the bottom conductors 4610b, 4620b, 4630b, 4640b is counterclockwise. While at each corner of the array, the vertical conductor currents of adjacent electromagnets (eg, a pair of vertical conductors 4620c and 4630d) cancel each other's magnetic field on the wafer surface. The net effect is to generate the clockwise and counterclockwise current loops at the top and bottom of the chamber, respectively, as in FIG. 44, resulting in the same cusp-type magnetic field as shown in FIG. is there. The reactor of FIG. 46 is operated in any of three modes:
A magnetic pressure mode in which a cusp-type magnetic field is generated;
A sinusoidal mode in which four sinusoidal currents are applied to four electromagnets 4610, 4620, 4630, 4640 in quadrature to generate a slow rotating magnetic field on the wafer surface;
Four electromagnets 4610, 4620, 4630, 4640 are grouped into opposing pairs of adjacent pairs, one pair having one DC current, the opposing pair having a reverse DC current, and the wafer This is a configurable magnetic field (CMF) mode that generates a substantially straight magnetic field line extending in a direction oblique to the direction of four electromagnets 4610, 4620, 4630, 4640 over the surface. This grouping is rotated by switching the current so that the magnetic field rotates through four diagonal directions. These orientation time sequences are illustrated in FIGS. 47A, 47B, 47C and 47D.

  [168] In FIG. 47A, electromagnets 4610, 4620 have a positive DC current, whereas electromagnets 4630, 4640 have a negative DC current, and the resulting average magnetic field direction is shown in FIG. From the upper left corner to the lower right corner. In FIG. 47B, the electromagnets 4620, 4630 have a positive current while the electromagnets 4610, 4640 have been switched to group so that they have a negative current, and the average magnetic field rotates 90 degrees clockwise. ing. 47C and 47D complete this cycle. The strength of the magnetic field line is determined by the difference between the magnitudes of the positive and negative DC currents thus applied and can be adjusted as desired by programming the controller 4650.

  [169] The method of FIG. 9 uses the CMF to accurately select the DC current of the four electromagnets 4610, 4620, 4630, 4640 to produce the best correction for non-uniform etch rate or plasma ion concentration distribution. Used in mode. When applying the method of FIG. 9 to the CMF mode of FIGS. 47A-D, the electromagnets or coils 4610, 4620, 4630, 4640 are used in place of the overhead coils 60, 65, and all steps of FIG. Follow that alternative. The only difference is that the calculation of the magnetic field from each coil is calculated as the average over the four periods corresponding to FIGS.

  [170] FIG. 48 illustrates a reactor including a special grid 4810 that is inserted over the pump ring. The grid 4810 is formed of a semiconductor material such as silicon carbide or a conductive material such as aluminum and has an opening 4820 that allows gas to escape from the chamber through the pump ring. A special grating 4810 eliminates the plasma from the pump ring and provides the necessary protection and process control. For this reason, the distance across the inside of each opening 4820 in the radial plane is about twice the plasma sheath thickness. Thus, if not impossible, it is very difficult for the plasma to penetrate the grating 4810. This reduces or eliminates the interaction of the plasma in the pump ring with the chamber.

  [171] FIGS. 49 and 50 illustrate an integrally formed removable chamber liner 4910 that incorporates the plasma constraining lattice 4810 of FIG. Liner 4910 covers the radially outer portion of the region above wafer 110 under chamber electrode 125. Thus, the liner 4910 includes an upper horizontal portion 4920 that covers the outer periphery of the chamber ceiling, a vertical portion 4930 that covers the chamber sidewalls, and a plasma constraining grating 4810 that includes not only the annular surface adjacent to the wafer 110 but also the pump ring. And a lower horizontal portion 4940 that covers it. In some cases, each of portions 4920, 4930, 4940 are formed together as monolithic silicon carbide pieces 4950. The liner 4910 further includes and is bonded to an aluminum base 4960 that is below the lower horizontal portion 4940 of the silicon carbide piece 4950. The aluminum base 4960 includes a relatively long and thin pair of downwardly extending annular rails 4962, 4964 that provides good electrical conduction to the grounded chamber structural elements below the wafer support pedestal 105.

  [172] The reactor can have temperature control elements 4972, 4974 in thermal contact with the downwardly extending annular rails 4962, 4964 and temperature control elements 4976 in thermal contact with the vertical side portion 4930. is there. The thermal control elements 4972, 4974, 4976 may include a cooling device including a coolant path and a heating device including an electric heater. It may be desirable to maintain the liner 4910 at a sufficiently high temperature (eg, on the order of 120 degrees F.) to minimize or prevent deposition of polymer or fluorocarbon compounds on the inner surface of the liner 4910.

  [173] Liner 4910 enhances process stability to provide a good ground return path. This is because the potential is uniform along the inner surface of the silicon carbide piece 4950 (including the surfaces facing the interior of the upper horizontal portion 4920, the vertical portion 4930, and the lower horizontal portion 4940). As a result, the liner 4910 provides power delivered from either the overhead electrode 125 or the wafer pedestal 105 on its interior facing surface to a uniform RF return path. One advantage is that plasma fluctuations concentrate the RF feedback current distribution at different sites on the inner surface of the liner 4910, and the impedance imparted to that current remains almost constant. This feature promotes process stability.

  [174] FIG. 51 illustrates the uniform processing of a rectangular semiconductor or dielectric workpiece 4910, such as a photolithography mask, where the overhead coils 60, 65 define a rectangular pattern symmetrical to the square pattern of the MERIE magnets 92, 94, 96, 98. FIG. 7 illustrates a variant in the case of FIG.

  [175] FIG. 52 illustrates a version in which the wafer support pedestal 105 of the reactor of FIG. 24 can move up and down. In addition to the two overhead coils 60, 65 that control the radial distribution of plasma ions, there is a bottom coil 5210 below the plane of the wafer support pedestal 105. In addition, there is an outer coil 5220 around the chamber. The outer overhead coil 65 and the bottom coil 5210 can have a reverse DC current to form a complete cusp-type magnetic field in the chamber.

  [176] Although the overhead coils 60, 65 have been described in combination with a reactor having an overhead ceiling that serves as both an overhead source power electrode and a gas distribution plate, this ceiling is not a gas distribution plate, but a process It may be of the type in which the gas is introduced in another conventional manner (eg through the side walls). Furthermore, coils 60 and 65 whose source power is not capacitively coupled by the sealing electrode can be used in the reactor. Further, an impedance matching element for an overhead electrode has been described as a fixed element such as a coaxial tuning stub. However, the impedance matching element may be any suitable or conventional impedance matching device, such as a conventional dynamic impedance matching circuit.

Three-magnet three-mode plasma distribution control:
[177] In plasma processes such as plasma enhanced reactive ion etching, a magnetic field is used to improve the uniformity of the radial distribution of etch rates across the semiconductor wafer. In most cases, the plasma ion concentration is high at the wafer center and low elsewhere on the wafer, so the etch rate tends to be high at the wafer center and low around the wafer. A magnetic field can be generated by the inner and outer coils 60, 65 to change the radial distribution of plasma ion concentration. Typically, the desired effect is to reduce the plasma ion concentration at the center and increase it at the wafer periphery. Inner and outer electromagnets (FIG. 1B) can be used to achieve such improved plasma ion concentration distribution uniformity. The magnetic field they generate together, such as radial component B _r (its flux lines are parallel to the planar wafer surface) and axial component B _z (its flux lines are orthogonal to the planar wafer surface), etc. It can be analyzed by breaking it down into two components. The radial component B _r of the magnetic field is most effective in changing or correcting the radial distribution of plasma ion concentration (eg, to achieve a uniform radial distribution of deposition in the etch process or CVD process). Is. However, changing the radial component B _r using only the inner and outer electromagnets 60, 65, necessarily changes the axial component B _z of the magnetic field as determined by the change in the radial component B _r. For example, an increase in B _r will typically cause avoidable impossible increase in B _z. This increase in such B _z is to not sought, it is undesirable. The inventors have discovered that control of the axial component B _z is closely associated with a reduction in device damage (eg, due to charge accumulation, high electric field, high induced current or voltage) on the wafer. The present inventors have found that in many cases, while simultaneously minimizing the axial component B _z, to optimize to the extent that the desired radial component B _r (e.g., increasing) the like.

  [178] Referring to FIG. 53A, the inner and outer electromagnets 60, 65 of FIG. 1B are augmented by a bottom electromagnet 401 just below the plane of the wafer. A DC current source 403 managed by the controller 90 provides a DC current to the bottom electromagnet 401, and current sources 70, 75 managed by the controller 90 provide a current to the inner and outer electromagnets 60, 65. Each electromagnet 60, 65, 401 in FIG. 53A consists of a single conductor winding, but may instead consist of a plurality of windings arranged in the vertical direction as shown in FIG. 53B. FIG. 54 illustrates how three electromagnets 60, 65, 401 can be placed in the reactor of FIG. As previously described in this specification, the reactor of FIG. 24 has an overhead VHF electrode that is driven by VHF plasma source power at a frequency at which the electrode and plasma resonate through a fixed impedance matching element.

[179] The magnets 60, 65, 401 can be used to generate any one (or combination) of three types of magnetic fields. (1) on the surface of the wafer 20 _{B r} and _{B z} of both a strong solenoid magnetic field (FIG. 55A). Such a solenoid magnetic field can be generated by applying a current to only one of the three electromagnets 60, 65, 401. FIG. 56A illustrates an embodiment in which a current is applied to the outer electromagnet 65 to generate a solenoid magnetic field. (2) A cusp magnetic field in which only _Br is strong and _Bz is weak or zero on the wafer surface (FIG. 55B). Such a cusp magnetic field can be generated by generating an equal and opposite magnetic field from the bottom electromagnet 401 and one of the inner and outer electromagnets 60,65. FIG. 56B illustrates an embodiment in which a reverse current is applied to the bottom and outer electromagnets 401, 65 to generate a cusp magnetic field. In this example, it is assumed that the coils of the magnets 65 and 401 are wound in the same direction (clockwise or counterclockwise). However, in the preferred embodiment, they are wound in the opposite direction, in which case the polarity of the applied current will be appropriately modified from that shown in FIG. (3) A mirror magnetic field that can be generated by generating a magnetic field in the same direction from the bottom electromagnet 401 and one of the inner and outer electromagnets 60 and 65 (FIG. 55C). FIG. 56C illustrates an embodiment in which equal currents are applied to the bottom and outer electromagnets 401, 65 to generate a mirror magnetic field.

[180] FIGS. 57A, 57B and 58A, 58B show the radial and axial magnetic field components B _r (r) and B of the cusp and mirror magnetic fields measured in the plane of the wafer 20 in the reactor of FIG. 53A. _z a (r) is compared with the radial and axial magnetic field components of the solenoid magnetic field _B r (r) and _B z (r). FIG. 57A compares the radial magnetic field component B _r (r) of the solenoid and the cusp magnetic field, and FIG. 57B compares the axial magnetic field component B _z (r) of the solenoid and the cusp magnetic field. 58A compares the radial magnetic field component B _r (r) of the solenoid and mirror magnetic fields, and FIG. 58B compares the axial magnetic field component B _z (r) of the solenoid and mirror magnetic fields. The radial components of the solenoid and cusp field can be approximately the same if desired (FIG. 57A), while the axial component of the cusp field is approximately zero (but not exactly zero) or the solenoid field. Much smaller than the axial component of (Fig. 57B). The radial component of the mirror field can be nearly zero if desired (FIG. 58A) or much smaller than that of the solenoid field. The axial component of the mirror field can be approximately the same as that of the solenoid field (FIG. 58B).

  [181] From FIGS. 57A and 58A, it can be seen that the radial magnetic field is likely to be ideal for correcting the central high plasma ion distribution as the radial magnetic field increases from the wafer center to the maximum intensity around the wafer. This is confirmed by the data depicted in the graph of FIG. 59 where the plasma ion concentration indicated by the plasma ion saturation current (vertical axis) is plotted as a function of the radius on the wafer surface (horizontal axis). The curve labeled “Zero Current” corresponds to zero magnetic field and indicates the minimum optimal placement for the central high plasma ion distribution. The best correction for the central high plasma ion distribution is two solenoid fields with either 5 amperes or 10 amperes in the external electromagnet 65 (labeled “5A solenoid” and “10A solenoid” respectively). According to the graph of FIG. 59, it is the solenoid magnetic field that increases most from the center of the wafer to the edge.

  [182] FIG. 60 is a graph of data obtained by measuring the etch rate for different magnetic fields as a function of radial position on the wafer surface of a 200 mm silicon wafer. The mirror field produces the best uniformity or lowest deviation of the etch rate distribution (deviation ratio of about 1.7%, where the deviation ratio is defined as the standard deviation divided by the average etch rate across the wafer). The best uniformity was then obtained with a solenoid field that produced a deviation ratio of about 2%. The cusp field (labeled “100% cusp”) is only the third best with a deviation ratio of about 7.9%. However, measurement of device damage on the wafer (due to charge accumulation, discharge or local high current or voltage conditions) yielded the opposite result. The most uniform case (mirror magnetic field) caused the most device damage, while the second most uniform case (solenoid magnetic field) had the second device damage, whereas the cusp magnetic field had little damage. These results will be described later with reference to FIG.

[183] The above results indicate that our findings that control over the radial component B _r (r) of the axial component B _z (r) of the magnetic field is closely related to improving device damage on the wafer. Is backed up. The cusp field produced little or no device damage. However, the inventors have a better prospect for the behavior of the radial component B _r (r) increasing with the radius shown in FIGS. 57A and 58A to reach more uniformity than the axial component B _z (r). I felt I gave it. Thus, the following method was implemented: As the wafer edge radial component B _r (r) is the same as the solenoid field of FIG. 60 that yielded such good results (ie, 22 Gauss at the wafer edge). The cusp magnetic field was adjusted. Next, the magnitude of the cusp field was increased (increasing B _r (r) while minimizing or zeroing B _z ) until a uniformity result close to the ideal result obtained with a solenoid field was obtained. . The present inventors have this finds it takes increased until the magnitude of the cusp field B _r of the wafer edge increased from 22 Gauss to 32 Gauss (or about 160%). This is the etching rate distribution curve indicated as “cusp 160%” in the graph of FIG. At that point, the etching rate deviation ratio was reduced to 2.4%. Despite its dramatic increase in strength, the cusp field still caused little or no device damage.

[184] The above is summarized in the table of FIG. The left column presents a magnetic field type, a wafer central B _z, describes a B _r at the wafer edge of the magnetic field in gauss. The middle column provides the corresponding etch rate deviation ratio (non-uniformity), the right column provides an assessment of device damage (“good” or “bad”), inductive device current (in milliamps) and voltage (in volts). ) Table of FIG. 61, good uniformity and poor device damage results obtained with the solenoid and mirror fields, and low uniformity obtained with the cusp field having a B _r at the wafer edge consistent with that of the solenoid magnetic field and Shows good device damage results. The last column of the table shows the good uniformity and good device damage results obtained when the cusp field (substantially free of other fields) was increased to its previous level of 160%. ing.

  [185] The above approach is facilitated by the method shown in FIG. The first step (block 415 in FIG. 62) determines the solenoid field strength that minimizes the radial distribution non-uniformity of the etch rate. This corresponds to the solenoid field of FIG. 60 having a radial component intensity of 22 Gauss at the wafer edge. Exact values can vary depending on the particular process. The radial component value at the selected radius (eg, wafer edge) is recorded (block 417). Next, a cusp field is established having the same radial component field strength at the selected radius as recorded in the step of block 417, assuming there are no other magnetic fields (block 419). Finally, the cusp field strength is increased until the etch rate radial distribution non-uniformity is minimized (block 421). This step corresponds to increasing the cusp radial component from 22 gauss to 32 gauss. However, these values can vary depending on the process being performed.

The cusp field in the method of FIG. 62 is established using the outer electromagnet 65 and the bottom electromagnet 401. Once the desired radial component B _r (r) is established, further trimming or correction can be performed by applying a relatively small current to the inner electromagnet 60 according to the method of FIG. The inner electromagnet current may be selected to further enhance uniformity or to control or reduce the axial component B _z (r) to improve device damage results (ie, reduce device damage). it can. This approach is implemented in the method of FIG. 63 and the first step (block 423) establishes the desired radial component strength B _r (r) using, for example, the method of FIG. Next, the condition is optimized (either further improving uniformity or removing B _z ) by applying a relatively small current to the inner electromagnet 60 (block 425 in FIG. 63).

[187] In a variation of this process illustrated in FIG. 64, the desired magnetic field (eg, ideal B _r (r)) is established with the inner and outer magnets 60, 65, and the bottom magnet is deactivated ( 64, block 431). Next, the magnetic field is trimmed at block 433 of FIG. 64 (eg, to increase _Br if desired) by increasing the current through the inner electromagnet 401 until the desired result is obtained. In some embodiments, improved plasma ion concentration distribution uniformity without actually causing an unacceptable increase in device damage by actually applying a very small axial magnetic field B _z from the inner electromagnet 60. Is obtained.

[188] In performing the process of FIG. 64, a desired set of DC current values for the inner and outer magnets 60, 65 for minimum etch rate distribution non-uniformity can be determined. This is accomplished by measuring the radial distribution of etch rates obtained for each value of the current of one of the inner and outer magnets 60, 65 when the other has a zero current. For example, FIG. 65 is a graph including a curve representing etching rate radial distribution data for different values of DC current applied to the inner magnet 60 within a selected range (0 amps to 25 amps). FIG. 66 is a graph including a curve representing the radial distribution data of the etching rate for different values of DC current applied to the outer magnet 65 within a selected range (0 amperes to 25 amperes). Overlaying different pairs of etch rate distribution curves from FIGS. 65 and 66 for a given pair of inner and outer magnet current values I _i , I _o until many or all of the possible pairs are juxtaposed, The resulting etch rate distribution and the corresponding etch rate radial distribution E (r) I _i, I _o obtained by superposition can be simulated. Each etch rate distribution is then processed to calculate a corresponding non-uniformity (eg, the deviation ratio D defined above). This produces a set of deviations D (I _i , I _o ) represented as a single surface shown in FIG. This surface or function can be examined using conventional techniques to determine a value or set of I _i , I _o that minimizes the deviation ratio D (vertical axis in FIG. 67). These are the values selected by the controller 90 for the inner and outer magnet currents.

[189] The above approach is implemented in the manner shown in FIG. First, the bottom magnet current is set to zero (block 435). The radial distribution of etch rates is measured for different inner magnet currents to obtain a set of distributions E (r) I _i (block 437), and different outer magnet currents to obtain a set of distributions E (r) _Io. Are measured (block 439). Two corresponding distribution pairs are superimposed to form different etch rate distributions E (r) I _i, I _o (block 441), from which the corresponding deviation D (I _i , I _o ) is calculated ( Block 443). A set of deviations D (I _i , I _o ) is represented by the surface (FIG. 67) and a set of values (I _i , I _o ) that results in the smallest deviation ratio D is searched (block 445).

[190] The inspection of the three-dimensional surface D (I _i , I _o ) of FIG. 67 corresponds to a series or list of consecutive optimal pairs (I _i , I _o ) where D (vertical axis) is minimized. Exposes a narrow valley (highlighted with a dashed line). This valley can be obtained by a conventional search. To optimize the use of the third magnet (ie bottom magnet 401), combining each of the optimal pairs (I _i , I _o ) with successive values of the bottom magnet current I _b in a predetermined range. Each combination of three currents (I _i , I _o , I _b ) is applied to the reactor and the etch rate deviation is measured. This last step is block 447 in FIG. This result can be interpolated to generate a set of deviation values D (I _i , I _o , I _b ) (block 449). These sets of values can be represented by a four-dimensional surface, and a set of values (I _i , I _o , I _b ) that minimizes D is searched using conventional techniques (block 451). This minimization can provide an improvement to the minimization obtained in the block 445 step using only two magnets. The final (I _i , I _o , I _b ) optimum values are applied to the respective electromagnets 60, 65, 401 during product wafer processing for optimum process uniformity.

  [191] The process of FIG. 68 can be summarized as follows. First, the characteristics of only one pair of the three magnets, for example the inner and outer magnets 60, 65, will be elucidated. This magnet pair is then treated as a single body and the simultaneous use of three magnets is optimized by characterizing with a third magnet, eg, bottom magnet 401. However, there are three possible orders to characterize the three magnets. One is the example given in FIG. Second, the first magnet pair to be characterized is the outer magnet 65 and the bottom magnet 401, and the third magnet is the inner magnet 60. Third, the first magnet pair to be characterized is the inner magnet 60 and the bottom magnet 401, and the third magnet is the outer magnet 65.

[192] FIG. 69 illustrates a second version of this process, with the first magnet pair being an outer magnet 65 and a bottom magnet 401 and the third magnet being an inner magnet 60. In the first step of FIG. 69, the inner magnet current is set to zero (block 435-1). A radial distribution of etch rates is measured for different bottom magnet currents to obtain a set of distributions E (r) I _b (block 437-1) and a different outer to obtain a set of distributions E (r) _Io. The magnet current is measured (block 439-1). Two corresponding distribution pairs are superimposed to form different etch rate distributions E (r) I _b, I _o (block 441-1), from which the corresponding deviation D (I _b , I _o ) is calculated. (Block 443-1). A set of deviations D (I _b , I _o ) is represented by the surface (same as in FIG. 67) and the set of values (I _i , I _o ) that results in the smallest deviation or deviation ratio D is searched (block 445-1).

[193] In order to optimize the use of the third magnet (ie, bottom magnet 401), each of the optimal pairs (I _i , I _o ) and a continuous value of bottom magnet current I _b within a predetermined range Each combination of the three currents (I _i , I _o , I _b ) is applied to the reactor and the etch rate deviation is measured. This last step is block 447-1 in FIG. This result can be interpolated to generate a set of deviation values D (I _i , I _o , I _b ) (block 449-1). These sets of values can be represented by a matrix (or a four-dimensional surface), and a set of values D (I _i , I _o , I _b ) is searched using conventional techniques that minimize the deviation or deviation ratio D (Block 451-1). The DC current applied to the three magnets 60, 65, 401 is established according to this final set of values.

[194] FIG. 70 is a flowchart illustrating another method for achieving a uniform plasma or etch rate distribution using three electromagnets 60, 65, 401. First, the nominal (uncorrected) etch rate distribution ER (r) is measured without applying current to the electromagnets 60, 65, 401 (block 461). Next, the change in etch rate radial distribution caused by the inner coil current I _i , ie, ΔER (r, I _i ), is measured for many different values of I _i (block 463). The change in etch rate radial distribution caused by the outer coil current I _o , ie, ΔER (r, I _o ), is measured for many different values of I _o (block 465). The change in etch rate radial distribution caused by the bottom coil current I _b , ie, ΔER (r, I _b ), is measured for many different values of I _b (block 467). Next, an etch rate distribution is calculated for each combination of different I _i , I _o, and I _b values (block 469).

ER (r, I _i , I _o , I _b ) = ER (r) + ΔER (r, I _i ) + ΔER (r, I _o ) + ΔER (r, I _b )

A non-uniformity or deviation or deviation ratio D (I _i , I _o , I _b ) for each of these distributions is calculated (block 471). The matrix D (I _i , I _o , I _b ) can be interpolated to provide a smooth function, and a set of values (I _i , I _o , I _b ) that minimize D is searched ( Block 473). An optimal set of DC currents (I _i , I _o , I _b ) is thus determined and applied to the three magnets 60, 65, 401 (block 475).

[195] FIGS. 71A-71E graphically illustrate one calculation guide for the etch rate distribution ER (r, I _i , I _o , I _b ). The nominal etch rate distribution ER (r) is depicted in the graph of FIG. 71A. The change ΔER (r, I _i ) from the nominal distribution caused by applying a 5 amp DC current to the inner electromagnet 60 is depicted in FIG. 71B. The change ΔER (r, I _o ) from the nominal distribution caused by applying a 1 amp DC current to the outer electromagnet 65 is depicted in FIG. 71C. The change ΔER (r, I _b ) from the nominal distribution caused by applying a 2 amp DC current to the bottom magnet is depicted in FIG. 71D. The sum of the etching rate distributions of FIGS. 71A to 71D is depicted in FIG. 71E, which is an etching rate distribution ER (r, I _i = 5, I _o = 1, I _b = 2).

[196] Another way to determine the optimal current (I _i , I _o , I _b ) of the three magnets is to use the etch rate distribution ER for many different combinations of values (I _i , I _o , I _b ). (R, I _i , I _o , I _b ) is directly measured. This approach involves a large number of measurements and replaces the steps of blocks 461-469 in FIG. Once a sufficient number of different ERs (r, I _i , I _o , I _b ) are thus measured, the steps of blocks 471, 473, and 475 of FIG. 70 are performed.

  [197] In the above process, uniformity was defined with reference to the radial distribution of etch rate across the wafer being etched in the reactor. More generally, however, for any process, including an etching process or a deposition process, process uniformity can also be defined as the uniformity of the radial distribution of plasma ion concentration across the wafer surface. In the etching reactor, the plasma ion concentration distribution is estimated from the etching rate radial distribution measured on the wafer processed in the plasma enhanced reactive ion etching process performed in the reactor.

  [198] While the reactor has been described in detail with reference to the preferred embodiments, it will be understood that modifications and variations can be made without departing from the spirit and scope of the reactor.

1 is an illustration of a plasma reactor with an overhead VHF electrode and an overhead coil for controlling plasma ion uniformity. 1 is an illustration of a plasma reactor with an overhead VHF electrode and an overhead coil for controlling plasma ion uniformity. 1 is an illustration of a plasma reactor with an overhead VHF electrode and an overhead coil for controlling plasma ion uniformity. FIG. 2 is an illustration of an exemplary apparatus for controlling the overhead coil of FIG. 1. 2 is a graphical representation of the magnetic field of the overhead coil of FIG. 2 is a graphical representation of the magnetic field of the overhead coil of FIG. It is a spatial representation of the same magnetic field. 2 is a graph of etch rate (vertical axis) on the wafer surface as a function of radial position (horizontal axis) for various modes of operation of the reactor of FIG. 2 is a graph of etch rate (vertical axis) on the wafer surface as a function of radial position (horizontal axis) for various modes of operation of the reactor of FIG. 2 is a graph of etch rate (vertical axis) on the wafer surface as a function of radial position (horizontal axis) for various modes of operation of the reactor of FIG. 2 is a graph of etch rate (vertical axis) on the wafer surface as a function of radial position (horizontal axis) for various modes of operation of the reactor of FIG. 2 is a graph of etch rate (vertical axis) on the wafer surface as a function of radial position (horizontal axis) for a further mode of operation of the reactor of FIG. 2 is a graph of etch rate (vertical axis) on the wafer surface as a function of radial position (horizontal axis) for a further mode of operation of the reactor of FIG. 2 is a graph of etch rate (vertical axis) on the wafer surface as a function of radial position (horizontal axis) for a further mode of operation of the reactor of FIG. 2 is a graph of etch rate (vertical axis) on the wafer surface as a function of radial position (horizontal axis) for a further mode of operation of the reactor of FIG. 3 is a graph depicting etching rate as a function of magnetic field. 1B is an illustration of the reactor of FIG. 1A with a MERIE magnet. 1B is an illustration of the reactor of FIG. 1A with a MERIE magnet. 1B depicts a method of operating the reactor of FIG. 1A. 1B is a graph illustrating a comparative example of magnetic pressure and ion or electron density as a function of radial position on the wafer surface in the reactor of FIG. 1A. 6 is a graph depicting non-uniformity in etch rate as a function of coil current. It is illustration of radial direction ion distribution in the zero coil current in the example of FIG. FIG. 12 is a diagram comparing the measured etching rate distribution with the predicted etching rate distribution at a coil current of about 11 amperes in the example of FIG. FIG. 12 is a diagram comparing the measured etching rate distribution with the predicted etching rate distribution at a coil current of about 11 amperes in the example of FIG. FIG. 12 compares the measured etch rate distribution with the predicted etch rate distribution at a coil current of about 35 amps in the example of FIG. FIG. 12 compares the measured etch rate distribution with the predicted etch rate distribution at a coil current of about 35 amps in the example of FIG. FIG. 1B depicts a further method of operating the reactor of FIG. 1A. 1B is an illustration of the magnetic field distribution obtained in the reactor corresponding to FIG. 1A. FIG. 17 depicts a gradient of the square of the magnetic field of FIG. 16 in the wafer plane. 1B is an illustration of another magnetic field distribution obtained in the reactor corresponding to FIG. 1A. FIG. 19 depicts a gradient of the square of the magnetic field of FIG. 18 in the wafer plane. 1B is an illustration of still further magnetic field distributions obtained in the reactor corresponding to FIG. 1A. FIG. 21 depicts the gradient of the square of the magnetic field of FIG. 20 in the wafer plane. FIG. 1B depicts another method of operating the reactor of FIG. 1A. 1B is an illustration of an exemplary microcontroller operation for controlling the reactor of FIG. 1A. 1B is an illustration of a plasma reactor including features included in the reactor of FIG. 1A. 1B is an illustration of another plasma reactor including features included in the reactor of FIG. 1A. 26 is an illustration of a gas distribution plate for the reactor of FIGS. 1A, 24 and 25. FIG. 26 is an illustration of a gas distribution plate for the reactor of FIGS. 1A, 24 and 25. FIG. 26 is an illustration of a gas distribution plate for the reactor of FIGS. 1A, 24 and 25. FIG. 26 is an illustration of a gas distribution plate for the reactor of FIGS. 1A, 24 and 25. FIG. 26 is an illustration of a gas distribution plate for the reactor of FIGS. 1A, 24 and 25. FIG. FIG. 27 is an illustration of thermal control features in the gas distribution plate as in FIG. 26. FIG. 27 is an illustration of thermal control features in the gas distribution plate as in FIG. 26. FIG. 27 is an illustration of a gas distribution plate with dual zone gas flow control corresponding to FIG. FIG. 27 is an illustration of a gas distribution plate with dual zone gas flow control corresponding to FIG. 1B is an illustration of a plasma reactor having a dual zone gas distribution plate corresponding to FIG. 1A. 1 is an illustration of an exemplary dual zone gas flow controller. 1 is an illustration of an exemplary dual zone gas flow controller. FIG. 35 is an illustration of a plasma reactor having three overhead coils for controlling the plasma ion distribution corresponding to FIG. FIG. 27 depicts different gas injection hole patterns that respectively generate a center low or center high gas flow distribution in the gas distribution plate of FIG. FIG. 27 depicts different gas injection hole patterns that respectively generate a center low or center high gas flow distribution in the gas distribution plate of FIG. FIG. 4 is an illustration of different arrangements of overhead coils for controlling plasma ion distribution. FIG. 4 is an illustration of different arrangements of overhead coils for controlling plasma ion distribution. FIG. 4 is an illustration of different arrangements of overhead coils for controlling plasma ion distribution. FIG. 4 is an illustration of different arrangements of overhead coils for controlling plasma ion distribution. FIG. 46 is an illustration of a plasma reactor corresponding to FIG. 1A in which the overhead coils have been replaced by upper or lower magnetic coils above and below the reactor chamber generating the cusp-type magnetic field best seen in FIG. FIG. 46 is an illustration of a plasma reactor corresponding to FIG. 1A in which the overhead coils have been replaced by upper or lower magnetic coils above and below the reactor chamber generating the cusp-type magnetic field best seen in FIG. FIG. 46 illustrates how the upper and lower coils of FIG. 44 can be replaced by a configurable magnetic field (CMF) coil operated in a manner to generate the cusp-type magnetic field of FIG. FIG. 47 is an illustration of the mode of operation of the CMF coil of FIG. 46 for generating a desired magnetic field setting. FIG. 47 is an illustration of the mode of operation of the CMF coil of FIG. 46 for generating a desired magnetic field setting. FIG. 47 is an illustration of the mode of operation of the CMF coil of FIG. 46 for generating a desired magnetic field setting. FIG. 47 is an illustration of the mode of operation of the CMF coil of FIG. 46 for generating a desired magnetic field setting. 1B is an illustration of an annular apertured plate for preventing plasma ions from entering the reactor pump ring in the reactor of FIG. 1B is an illustration of an annular apertured plate for preventing plasma ions from entering the reactor pump ring in the reactor of FIG. 1B is an illustration of an annular apertured plate for preventing plasma ions from entering the reactor pump ring in the reactor of FIG. 1B is an illustration of a rectangular version of the reactor of FIG. 1A for processing a rectangular workpiece. 1B is an illustration of a reactor corresponding to FIG. 1A having a retractable workpiece support pedestal. FIG. 6 is an illustration of different embodiments using two overhead coils and one underlying coil to control plasma ion distribution. FIG. 6 is an illustration of different embodiments using two overhead coils and one underlying coil to control plasma ion distribution. It is the figure on which one Embodiment of this invention was drawn. FIG. 55 depicts three magnetic fields at the wafer plane corresponding to the three modes of the reactor of FIG. 54. FIG. 55 depicts three magnetic fields at the wafer plane corresponding to the three modes of the reactor of FIG. 54. FIG. 55 depicts three magnetic fields at the wafer plane corresponding to the three modes of the reactor of FIG. 54. FIG. 56 is a diagram depicting applied electromagnet DC currents corresponding to FIGS. 55A-55C, respectively. FIG. 56 is a diagram depicting applied electromagnet DC currents corresponding to FIGS. 55A-55C, respectively. FIG. 56 is a diagram depicting applied electromagnet DC currents corresponding to FIGS. 55A-55C, respectively. It is a graph which compares carefully the distribution of the radial direction component and axial direction component of the cusp mode of the reactor of FIG. 54, and a solenoid mode. It is a graph which compares carefully the distribution of the radial direction component and axial direction component of the cusp mode of the reactor of FIG. 54, and a solenoid mode. It is a graph which compares carefully the distribution of the radial direction component of the solenoid mode of a reactor of FIG. 54, and a mirror mode, and an axial direction component. It is a graph which compares carefully the distribution of the radial direction component of the solenoid mode of a reactor of FIG. 54, and a mirror mode, and an axial direction component. FIG. 56 is a graph of plasma ion radial distribution (inferred from ion saturation current) for different modes of the reactor of FIG. FIG. 56 is a graph of etch rate as a function of radius for different magnetic fields or modes of the reactor of FIG. FIG. 5 is a table that reveals the identification of different magnetic fields or modes depending on etch rate non-uniformity and device damage. FIG. 6 depicts the basic process for determining the optimal DC coil current for at least two of the three magnets. FIG. 63 depicts an additional process that can be continued to the process of FIG. 62 to determine the optimal DC coil current for all three magnets. FIG. 64 depicts an alternative to the process of FIG. It is the graph which plotted the radial direction distribution of the etching rate about different magnetic field strength according to the 1st search with one magnet. FIG. 6 is a graph depicting the radial distribution of etch rates for different magnetic field strengths according to a second search with another magnet. FIG. 67 is a graph depicting a mathematical distribution function established from the distributions of FIGS. 65 and 66. FIG. FIG. 6 depicts one process for determining an optimal electromagnet DC current. FIG. 6 depicts another process for determining an optimal electromagnet DC current. FIG. 6 depicts yet another process for determining an optimal electromagnet DC current. FIG. 71 depicts an etch rate distribution obtained in successive steps of the process of FIG. 70. FIG. 71 depicts an etch rate distribution obtained in successive steps of the process of FIG. 70. FIG. 71 depicts an etch rate distribution obtained in successive steps of the process of FIG. 70. FIG. 71 depicts an etch rate distribution obtained in successive steps of the process of FIG. 70. FIG. 71 depicts an etch rate distribution obtained in successive steps of the process of FIG. 70.

Explanation of symbols

DESCRIPTION OF SYMBOLS 5 ... Cylindrical side wall, 10 ... Sealing, 15 ... Wafer support pedestal, 20 ... Wafer, 25 ... Process gas supply source, 40 ... RF generator, 60 ... Inner coil, 65 ... outer coil, 90 ... controller, 70, 75, 76 ... DC current supply source, 92, 94, 96, 98 ... MERIE electromagnet, 99 ... MERIE current controller.

Claims (20)

A plasma reactor for processing a workpiece,
A vacuum chamber with side walls and a sealing;
A workpiece support pedestal having a workpiece support surface in the chamber, facing the ceiling and comprising a cathode electrode;
An RF power generator coupled to the cathode electrode;
An outer annular inner electromagnet in a first plane overlying the workpiece support surface;
An outer annular outer electromagnet having a larger diameter than the inner electromagnet in a second plane overlying the workpiece support surface;
An external annular bottom electromagnets in the third plane underlying the workpiece support surface,
An inner, outer and bottom DC current source connected to each of the inner, outer and bottom electromagnets;
A processor for controlling DC current from the inner, outer and bottom DC current sources;
With
The processor is
(A) a cusp mode in which the DC current causes the bottom electromagnet and one of the inner and outer electromagnets to generate an equal and opposite magnetic field on the workpiece support surface;
(B) a mirror mode in which the DC current causes the bottom electromagnet and one of the inner and outer electromagnets to generate a magnetic field in the same direction on the workpiece support surface;
(C) the DC current is operable in three modes comprising a solenoid mode that causes at least one of the electromagnets to generate both radial and axial magnetic fields at the workpiece support surface;
Plasma reactor. It said workpiece support pedestal and said inner, outer and bottom electromagnets are coaxial plasma reactor according to claim 1.   The plasma reactor according to claim 2, wherein the first plane is above the second plane, and both the first plane and the second plane are above the third plane.   The plasma reactor of claim 3, wherein the first, second, and third planes are parallel to the workpiece support surface. The plasma reactor of claim 1 , wherein the processor is operable in only one of the three modes at a time. The plasma reactor of claim 1 , wherein the processor is operable in a selected one of the three modes. A plasma reactor for processing a workpiece,
A vacuum chamber with side walls and a sealing;
A workpiece support pedestal having a workpiece support surface in the chamber, facing the ceiling and comprising a cathode electrode;
An RF power generator coupled to the cathode electrode;
An outer annular inner electromagnet in a first plane overlying the workpiece support surface;
An outer annular outer electromagnet having a larger diameter than the inner electromagnet in a second plane overlying the workpiece support surface;
An outer annular bottom electromagnet in a third plane below the workpiece support surface;
An inner, outer and bottom DC current source connected to each of the inner, outer and bottom electromagnets;
A processor for controlling DC current from the inner, outer and bottom DC current sources;
With
The processor is
(A) a cusp mode in which the bottom electromagnet and one of the inner and outer electromagnets mainly generate a radial DC magnetic field;
(B) a mirror mode in which the bottom electromagnet and one of the inner and outer electromagnets mainly generate an axial magnetic field;
(C) A plasma reactor in which one of the electromagnets is operable in three modes including a solenoid mode that generates axial and radial magnetic fields. The plasma reactor of claim 7 , wherein the processor is operable in a selected one of the three modes. The plasma reactor of claim 7 , wherein the processor is operable to generate the three modes of components simultaneously. An outer annular inner electromagnet in a first plane overlying the workpiece support surface; and an outer annular outer electromagnet having a larger diameter than the inner electromagnet in a second plane above the workpiece support surface; , plasma reaction having an external annular bottom electromagnets in the third plane underlying the workpiece support surface, the inner, inside which are connected to the respective outer and bottom electromagnets, and an outer and bottom DC current source A method for improving the uniformity of plasma ion concentration distribution in a vessel, comprising:
Sufficient to increase the plasma ion concentration near the periphery of the workpiece support surface relative to the plasma ion concentration at the center of the workpiece support surface from the bottom electromagnet and one of the inner and outer electromagnets. Generating a radial magnetic field having a magnetic field strength at the workpiece support surface;
Prior to processing the product workpiece, further comprising determining a solenoid magnetic field that produces a desired uniformity of plasma ion concentration radial distribution and determining a radial component of said solenoid magnetic field;
The step of generating the radial magnetic field comprises converting the radial magnetic field into the radial component of the solenoid magnetic field until the uniformity of the plasma ion concentration radial distribution reaches the desired uniformity generated by the solenoid magnetic field. Said method comprising increasing beyond intensity . The method of claim 10 , further comprising further increasing plasma ion concentration at the periphery by generating an additional magnetic field component at the other of the inner and outer electromagnets. The method of claim 10 , wherein the additional magnetic field component comprises an axial magnetic field at the workpiece support surface. The axial magnetic field has a lower magnetic field strength at the workpiece support surface than the radial magnetic field;
The method of claim 12 . An outer annular inner electromagnet in a first plane overlying the workpiece support surface; and an outer annular outer electromagnet having a larger diameter than the inner electromagnet in a second plane above the workpiece support surface; in the plasma reactor having an external annular bottom electromagnets in the third plane underlying the workpiece support surface, a method of controlling plasma ion density distribution,
A step tends to minimize the non-uniformity of plasma ion density distribution, obtaining a set of inner, DC current vs. applied to a pair of outer and bottom electromagnets,
For each of said DC current to tend to minimize the non-uniformity of plasma ion density distribution, the inner, seeking DC current applied to the other of the outer and bottom electromagnets, said inner, outer and bottom establishing a set of three DC current corresponding to the electromagnets,
One of the inner side of the three DC current, comprising the steps of applying to the outer and bottom electromagnets, plasma ion density distribution, the workpiece diameter measured etch rate on the processed semiconductor wafers in which the support surface The step estimated from the directional distribution;
A method comprising: Inside, it said pair of outer and bottom electromagnets and a one of said inner and outer electromagnets and said bottom electromagnet, whereby the electromagnet pair is established predominantly radial magnetic field at said workpiece support surface, the other electromagnets 15. The method of claim 14 , wherein establishes a smaller axial magnetic field. Inside, it said pair of outer and bottom electromagnets is provided with the bottom electromagnets and outer electromagnets, the other electromagnets comprises the inner electromagnets method of claim 15. An outer annular inner electromagnet in a first plane overlying the workpiece support surface; and an outer annular outer electromagnet having a larger diameter than the inner electromagnet in a second plane above the workpiece support surface; in the plasma reactor having an external annular bottom electromagnets in the third plane underlying the workpiece support surface, a method of controlling plasma ion density distribution, uncorrected by said workpiece support surface Determining a plasma ion concentration distribution;
Determining a change in plasma ion density distribution as a function of the inner, independent DC current applied one to solely the outer and bottom electromagnets,
Said inner, different combinations of DC current applied to the outer and bottom electromagnets, said function superimposed on said uncorrected plasma distribution, and obtaining a plurality of pilot plasma ion density distribution,
Searching the experimental plasma ion concentration distribution for at least one of the plasma ion concentration distributions having high uniformity and determining an optimal set of currents corresponding thereto;
And applying the optimal set of current in each of said inner, outer and bottom electromagnets,
With
The method wherein the step of determining the plasma ion concentration distribution comprises inferring the plasma ion concentration distribution from an etch rate distribution measured on a semiconductor wafer processed on the workpiece support surface . A plasma reactor for processing a workpiece on a workpiece support surface in a reactor chamber comprising:
An outer annular inner electromagnet in a first plane overlying the workpiece support surface;
An outer annular outer electromagnet having a larger diameter than the inner electromagnet in a second plane overlying the workpiece support surface;
An external annular bottom electromagnets in the third plane underlying the workpiece support surface;
A processor for controlling a DC current applied to each of the inner, outer and bottom electromagnets;
A memory accessible to the processor, wherein the memory stores a value of a DC current for each of the inner, outer and bottom electromagnets;
Determining an uncorrected plasma ion concentration distribution at the workpiece support surface;
Determining a change in plasma ion density distribution as a function of the inner, independent DC current applied one to solely the outer and bottom electromagnets,
Said inner, different combinations of DC current applied to the outer and bottom electromagnets, said function superimposed on said uncorrected plasma distribution, to obtain a plurality of trial plasma ion density distribution,
Searching the experimental plasma ion concentration distribution for at least one of the plasma ion concentration distributions having high uniformity and determining an optimal set of currents corresponding thereto;
Said memory determined by a process comprising:
With
The plasma reactor wherein the plasma ion concentration distribution is inferred from an etch rate distribution measured on a wafer processed on the workpiece support surface . The ceiling comprises a capacitively coupled overhead electrode, and the plasma reactor comprises:
A VHF plasma source power generator;
A fixed tuning element coupling the VHF plasma source power generator to the overhead electrode;
The electrode forming resonance with the plasma in the chamber having a resonance frequency at or near the frequency of the VHF plasma source power generator;
The plasma reactor according to claim 1, further comprising: The plasma reactor of claim 19 , wherein the fixed tuning element comprises a coaxial tuning stub having a stub resonance frequency at or near the resonance frequency.

Images